Version 1.21RC5

New features:
- New M569 T parameter options to specify step pulse width, step pulse
interval, direction setup and direction hold times
- M665 now sets the M208 limits (except Z min) to match the machine
limits, so that Duet Web Control reports the correct values

Bug fixes:
- M116 commands were sometimes executed out-of-order relative to
previous G10 commands if movement commands were in progress
- G10 L20 now computes the workplace coordinate origin correctly
- G53 is only active until the end of the current line of GCode
- The default coordinate system is the one selected by G54
- G54..G59 update the user coordinates immediately
- Fixed "Error: Pop(): stack underflow!" when a file or macro is
terminated due to an illegal move command
- A short delay is inserted when M558 is used to change the Z probe
type, to allow the averaging filters to accumulate the new data
- M291 messages which are non-blocking (i.e. mode < 2) are now
synchronised to queued moves, like M117 messages

Other changes:
- Duet Web Control and PanelDue now report the coordinates relative to
the origin of the current workplace
- Rewrote dhcp_rec function on Duet06/085 to avoid goto statements
(possible fix for startup problem when using DHCP)

23 files changed, 346 insertions, 170 deletions

diff --git a/src/BugList.txt b/src/BugList.txt
index 0afe7b67..5916a9b5 100644
--- a/src/BugList.txt
+++ b/src/BugList.txt
@@ -60,31 +60,45 @@ Done in 1.21RC2:
 - [done, should be better] https://www.duet3d.com/forum/thread.php?pid=40183#p40183, https://www.duet3d.com/forum/thread.php?pid=39509#p39509 (not starting up correctly in access point mode)
 - [done] Attempted movements outside limits or when not homed cause current print file to be aborted
 
-Remaining:
+Done in 1.21RC4:
 - [done] for CoreXZ require just X and Y to be homed before probing (so that G30 can be used to home Z axis)
-- [done, test] Check min. 8 character password in M589, see https://www.duet3d.com/forum/thread.php?pid=40914#p40914
 - [done] G10 L2
 - [done] G10 L20
 - [done, test] Suppress heater fault detection when heaters are suspended
 - [done] DuetWiFiServer: new SDK version
 - [problem gone away with latest DuetWiFiServer] Duet WiFi (Kossel): disconnects every time after I upload file xyzCalibration_cube s3d.gcode due to incomplete json reply, but can reconnect afterwards
-- [done] Ethernet: set W5500 to auto negotiate
+- [done, ok] Ethernet: set W5500 to auto negotiate
+- [done, test] Check min. 8 character password in M589, see https://www.duet3d.com/forum/thread.php?pid=40914#p40914
 - [done, test] I parameter on M452, M453 and M573 didn't work
 - [done, test] M564 H0 to allow axis movement before homing on Cartesian/CoreXY printers
 - [done, test] Filament length comment generated by Cura, see https://www.duet3d.com/forum/thread.php?pid=40913#p40913
 - [done, test] Doesn't cancel homeall.g if there is an illegal Z move at the start, https://www.duet3d.com/forum/thread.php?pid=41082#p41082
+- [done] Corrected Z probe input pin in RADDS build
+- [done, test] in M589 check password is empty or at least 8 characters
+- [done] Multi tap on G30
+- [done, didn't help] Possible fix to Duet 06/085 failure to start when using DHCP
+- [ok] Test smart effector sensitivity programming
 
+Remaining:
+- [done] M116 fix, see https://www.duet3d.com/forum/thread.php?pid=41954#p41954. Also M109, M190, M191
+- [done] queue M291 non-blocking messages
+- [done, networking seems OK inc. DHCP] Duet 06/085 DHCP issue; rewrite dhcp_recv without goto? https://www.duet3d.com/forum/thread.php?pid=34605#p34605 and https://www.duet3d.com/forum/thread.php?pid=41923#p41923
+- [done, hard to test] Error: Pop(): stack underflow! https://www.duet3d.com/forum/thread.php?pid=41967#p41967
+- [done, test] Short delay after M558, https://www.duet3d.com/forum/thread.php?pid=41967#p41967
+- [done, test] G10 L20 sign error, https://www.duet3d.com/forum/thread.php?pid=42044#p42044
+- [done, test] When G54 etc. are sent, update user coordinates immediately
+- [done, test] G54 is the default coordinate system; G53 only lasts until end of line
+- [done, test] DWC and PD report user coordinates, nt machine coordinates
+- [done] new step pulse timing for slow drivers
+- check out G30 H parameter, https://www.duet3d.com/forum/thread.php?pid=42322#p42322
+
+- Error message if you attempt movement with VIN < minimum
 - How should we lift X on a CoreXZ printer before homing?
-- Multi tap on G30?
-
 - if a homing command in an SD print file is aborted due to e.g. G1 Z5 in the homing file, error message should be written to both DWC and PanelDue
-- M116 issue
-- [made a change, but try to reproduce it] Duet 0.8/0.8.5 DHCP issue, https://www.duet3d.com/forum/thread.php?pid=34605#p34605
 - stall detect on Z axis
 - Files generated by Cura doesn't detect layer changes, see https://www.duet3d.com/forum/thread.php?pid=40865#p40865
 - [re-test using new DuetWiFiServer] "Failed to change mode" messages after M552 S2/S0/S1 cycle
 - [can't reproduce] "Attempt to seek on a non-open file", https://www.duet3d.com/forum/thread.php?pid=41175#p41175
-- Test smart effector sensitivity programming
 
 - min/max RSSI display?
 
diff --git a/src/Duet/Lwip/lwip/src/core/dhcp.c b/src/Duet/Lwip/lwip/src/core/dhcp.c
index 3b9dea30..f5483bc7 100644
--- a/src/Duet/Lwip/lwip/src/core/dhcp.c
+++ b/src/Duet/Lwip/lwip/src/core/dhcp.c
@@ -1568,12 +1568,24 @@ dhcp_recv(void *arg, struct udp_pcb *pcb, struct pbuf *p, ip_addr_t *addr, u16_t
 
   if (p->len < DHCP_MIN_REPLY_LEN) {
     LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_WARNING, ("DHCP reply message or pbuf too short\n"));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
     goto free_pbuf_and_return;
+#endif
   }
 
   if (reply_msg->op != DHCP_BOOTREPLY) {
     LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_WARNING, ("not a DHCP reply message, but type %"U16_F"\n", (u16_t)reply_msg->op));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
     goto free_pbuf_and_return;
+#endif
   }
   /* iterate through hardware address and match against DHCP message */
 #if 1 //dc42 from lwip2
@@ -1585,27 +1597,51 @@ dhcp_recv(void *arg, struct udp_pcb *pcb, struct pbuf *p, ip_addr_t *addr, u16_t
       LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_WARNING,
         ("netif->hwaddr[%"U16_F"]==%02"X16_F" != reply_msg->chaddr[%"U16_F"]==%02"X16_F"\n",
         (u16_t)i, (u16_t)netif->hwaddr[i], (u16_t)i, (u16_t)reply_msg->chaddr[i]));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
       goto free_pbuf_and_return;
+#endif
     }
   }
   /* match transaction ID against what we expected */
   if (ntohl(reply_msg->xid) != dhcp->xid) {
     LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_WARNING,
       ("transaction id mismatch reply_msg->xid(%"X32_F")!=dhcp->xid(%"X32_F")\n",ntohl(reply_msg->xid),dhcp->xid));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
     goto free_pbuf_and_return;
+#endif
   }
   /* option fields could be unfold? */
   if (dhcp_parse_reply(dhcp, p) != ERR_OK) {
     LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_SERIOUS,
       ("problem unfolding DHCP message - too short on memory?\n"));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
     goto free_pbuf_and_return;
+#endif
   }
 
   LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE, ("searching DHCP_OPTION_MESSAGE_TYPE\n"));
   /* obtain pointer to DHCP message type */
   if (!dhcp_option_given(dhcp, DHCP_OPTION_IDX_MSG_TYPE)) {
     LWIP_DEBUGF(DHCP_DEBUG | LWIP_DBG_TRACE | LWIP_DBG_LEVEL_WARNING, ("DHCP_OPTION_MESSAGE_TYPE option not found\n"));
+#if 1	//dc42
+    dhcp->msg_in = NULL;
+    pbuf_free(p);
+    return;
+#else
     goto free_pbuf_and_return;
+#endif
   }
 
   /* read DHCP message type */
@@ -1653,14 +1689,10 @@ dhcp_recv(void *arg, struct udp_pcb *pcb, struct pbuf *p, ip_addr_t *addr, u16_t
     /* remember offered lease */
     dhcp_handle_offer(netif);
   }
+#if 0	//dc42
 free_pbuf_and_return:
-#if 1 //dc42 from lwpi2
-if (dhcp != NULL) {
-  dhcp->msg_in = NULL;
-}
-#else
-  dhcp->msg_in = NULL;
 #endif
+  dhcp->msg_in = NULL;
   pbuf_free(p);
 }
 
diff --git a/src/DuetNG/SX1509.h b/src/DuetNG/SX1509.h
index 76a3a92a..96a6a59b 100644
--- a/src/DuetNG/SX1509.h
+++ b/src/DuetNG/SX1509.h
@@ -27,7 +27,7 @@ Distributed as-is; no warranty is given.
 #include "Core.h"
 
 const int ReceiveTimeout = 1000;			// Timeout for I2C receive
-const uint8_t DefaultOscDivider = 5;		// a clock divider of 2 ^ (5 - 1) = 16 gives a PSWM frequency of 2MHz / (16 * 255) = 488Hz
+const uint8_t DefaultOscDivider = 5;		// a clock divider of 2 ^ (5 - 1) = 16 gives a PWM frequency of 2MHz / (16 * 255) = 488Hz
 
 class SX1509
 {
diff --git a/src/Fan.cpp b/src/Fan.cpp
index 0a3deb9f..367433b0 100644
--- a/src/Fan.cpp
+++ b/src/Fan.cpp
@@ -116,7 +116,7 @@ bool Fan::Configure(unsigned int mcode, int fanNum, GCodeBuffer& gb, const Strin
 			}
 			if (heatersMonitored != 0)
 			{
-				SetValue(1.0);			// default the fan speed to full for safety
+				SetPwm(1.0);			// default the fan speed to full for safety
 			}
 		}
 
@@ -124,7 +124,7 @@ bool Fan::Configure(unsigned int mcode, int fanNum, GCodeBuffer& gb, const Strin
 		if (seen && gb.Seen('S'))		// Set new fan value - process this after processing 'H' or it may not be acted on
 		{
 			const float f = constrain<float>(gb.GetFValue(), 0.0, 255.0);
-			SetValue(f);
+			SetPwm(f);
 		}
 
 		if (seen)
@@ -158,7 +158,7 @@ bool Fan::Configure(unsigned int mcode, int fanNum, GCodeBuffer& gb, const Strin
 	return seen;
 }
 
-void Fan::SetValue(float speed)
+void Fan::SetPwm(float speed)
 {
 	if (speed > 1.0)
 	{
diff --git a/src/Fan.h b/src/Fan.h
index 19cbd1cd..9fb9d9ad 100644
--- a/src/Fan.h
+++ b/src/Fan.h
@@ -25,10 +25,10 @@ public:
 	bool Configure(unsigned int mcode, int fanNum, GCodeBuffer& gb, const StringRef& reply, bool& error);
 
 	bool IsEnabled() const { return pin != NoPin; }
-	float GetValue() const { return val; }
+	float GetConfiguredPwm() const { return val; }			// returns the configured PWM. Actual PWM may be different, e.g. due to blipping or for thermostatic fans.
 
 	void Init(Pin p_pin, bool hwInverted);
-	void SetValue(float speed);
+	void SetPwm(float speed);
 	void SetHeatersMonitored(HeatersMonitoredBitmap h);
 	bool Check();											// update the fan PWM returning true if it is a thermostatic fan that is on
 	void Disable();
diff --git a/src/GCodes/GCodeBuffer.cpp b/src/GCodes/GCodeBuffer.cpp
index ccb20d82..a3bc359b 100644
--- a/src/GCodes/GCodeBuffer.cpp
+++ b/src/GCodes/GCodeBuffer.cpp
@@ -390,6 +390,7 @@ void GCodeBuffer::SetFinished(bool f)
 		}
 		else
 		{
+			machineState->useMachineCoordinates = false;		// G53 does not persist beyond the current line
 			Init();
 		}
 	}
@@ -445,13 +446,14 @@ float GCodeBuffer::GetFValue()
 }
 
 // Get a colon-separated list of floats after a key letter
+// If doPad is true then we allow just one element to be given, in which case we fill all elements with that value
 const void GCodeBuffer::GetFloatArray(float arr[], size_t& returnedLength, bool doPad)
 {
 	if (readPointer >= 0)
 	{
 		size_t length = 0;
 		bool inList = true;
-		while(inList)
+		while (inList)
 		{
 			if (length >= returnedLength)		// array limit has been set in here
 			{
@@ -475,7 +477,7 @@ const void GCodeBuffer::GetFloatArray(float arr[], size_t& returnedLength, bool
 		// Special case if there is one entry and returnedLength requests several. Fill the array with the first entry.
 		if (doPad && length == 1 && returnedLength > 1)
 		{
-			for(size_t i = 1; i < returnedLength; i++)
+			for (size_t i = 1; i < returnedLength; i++)
 			{
 				arr[i] = arr[0];
 			}
@@ -930,6 +932,7 @@ bool GCodeBuffer::PushState()
 	ms->runningM501 = machineState->runningM501;
 	ms->runningM502 = machineState->runningM502;
 	ms->volumetricExtrusion = false;
+	ms->useMachineCoordinates = false;
 	ms->messageAcknowledged = false;
 	ms->waitingForAcknowledgement = false;
 	machineState = ms;
diff --git a/src/GCodes/GCodeMachineState.cpp b/src/GCodes/GCodeMachineState.cpp
index dc396a3d..696d8f07 100644
--- a/src/GCodes/GCodeMachineState.cpp
+++ b/src/GCodes/GCodeMachineState.cpp
@@ -13,7 +13,8 @@ unsigned int GCodeMachineState::numAllocated = 0;
 // Create a default initialised GCodeMachineState
 GCodeMachineState::GCodeMachineState()
 	: previous(nullptr), feedrate(DefaultFeedrate * SecondsToMinutes), fileState(), lockedResources(0), errorMessage(nullptr), state(GCodeState::normal),
-	  drivesRelative(false), axesRelative(false), doingFileMacro(false), runningM501(false), runningM502(false), volumetricExtrusion(false), waitingForAcknowledgement(false), messageAcknowledged(false)
+	  drivesRelative(false), axesRelative(false), doingFileMacro(false), runningM501(false), runningM502(false),
+	  volumetricExtrusion(false), useMachineCoordinates(false), waitingForAcknowledgement(false), messageAcknowledged(false)
 {
 }
 
diff --git a/src/GCodes/GCodeMachineState.h b/src/GCodes/GCodeMachineState.h
index 9ac29a64..d1455986 100644
--- a/src/GCodes/GCodeMachineState.h
+++ b/src/GCodes/GCodeMachineState.h
@@ -104,6 +104,7 @@ public:
 		runningM501 : 1,
 		runningM502 : 1,
 		volumetricExtrusion : 1,
+		useMachineCoordinates : 1,		// true if seen G53 on this line of GCode
 		// Caution: these next 3 will be modified out-of-process when we use RTOS, so they will need to be individual bool variables
 		waitingForAcknowledgement : 1,
 		messageAcknowledged : 1,
diff --git a/src/GCodes/GCodeQueue.cpp b/src/GCodes/GCodeQueue.cpp
index 0aa44556..207bfd17 100644
--- a/src/GCodes/GCodeQueue.cpp
+++ b/src/GCodes/GCodeQueue.cpp
@@ -60,6 +60,14 @@ GCodeQueue::GCodeQueue() : freeItems(nullptr), queuedItems(nullptr)
 			case 420:	// set RGB colour
 				return true;
 
+			case 291:
+				{
+					bool seen = false;
+					int32_t sParam = 1;
+					gb.TryGetIValue('S', sParam, seen);
+					return sParam < 2;					// queue non-blocking messages only
+				}
+
 			default:
 				break;
 			}
diff --git a/src/GCodes/GCodes.cpp b/src/GCodes/GCodes.cpp
index ac9d9c0c..6062b9e8 100644
--- a/src/GCodes/GCodes.cpp
+++ b/src/GCodes/GCodes.cpp
@@ -160,7 +160,7 @@ void GCodes::Reset()
 	{
 		axisScaleFactors[i] = 1.0;
 #if SUPPORT_WORKPLACE_COORDINATES
-		for (size_t j = 0; j < 10; ++j)
+		for (size_t j = 0; j < NumCoordinateSystems; ++j)
 		{
 			workplaceCoordinates[j][i] = 0.0;
 		}
@@ -320,7 +320,7 @@ void GCodes::Spin()
 		if (gb.MachineState().messageAcknowledged)
 		{
 			const bool wasCancelled = gb.MachineState().messageCancelled;
-			Pop(gb);
+			gb.PopState();											// this could fail if the current macro has already been aborted
 
 			if (wasCancelled)
 			{
@@ -527,7 +527,7 @@ void GCodes::RunStateMachine(GCodeBuffer& gb, const StringRef& reply)
 		if (LockMovementAndWaitForStandstill(gb))		// wait for tpre.g to finish executing
 		{
 			reprap.SelectTool(gb.MachineState().newToolNumber, simulationMode != 0);
-			GetCurrentUserPosition();					// get the actual position of the new tool
+			UpdateCurrentUserPosition();					// get the actual position of the new tool
 
 			gb.AdvanceState();
 			if (AllAxesAreHomed())
@@ -2347,6 +2347,12 @@ const char* GCodes::DoStraightMove(GCodeBuffer& gb, bool isCoordinated)
 			{
 				currentUserPosition[axis] += moveArg;
 			}
+#if SUPPORT_WORKPLACE_COORDINATES
+			else if (gb.MachineState().useMachineCoordinates)
+			{
+				currentUserPosition[axis] = moveArg - workplaceCoordinates[currentCoordinateSystem][axis];
+			}
+#endif
 			else
 			{
 				currentUserPosition[axis] = moveArg;
@@ -2475,6 +2481,13 @@ const char* GCodes::DoArcMove(GCodeBuffer& gb, bool clockwise)
 		currentUserPosition[X_AXIS] += xParam;
 		currentUserPosition[Y_AXIS] += yParam;
 	}
+#if SUPPORT_WORKPLACE_COORDINATES
+	else if (gb.MachineState().useMachineCoordinates)
+	{
+		currentUserPosition[X_AXIS] = xParam - workplaceCoordinates[currentCoordinateSystem][X_AXIS];
+		currentUserPosition[Y_AXIS] = yParam - workplaceCoordinates[currentCoordinateSystem][Y_AXIS];
+	}
+#endif
 	else
 	{
 		currentUserPosition[X_AXIS] = xParam;
@@ -4231,7 +4244,7 @@ void GCodes::SetMachinePosition(const float positionNow[DRIVES], bool doBedCompe
 }
 
 // Get the current position from the Move class
-void GCodes::GetCurrentUserPosition()
+void GCodes::UpdateCurrentUserPosition()
 {
 	reprap.GetMove().GetCurrentUserPosition(moveBuffer.coords, 0, reprap.GetCurrentXAxes(), reprap.GetCurrentYAxes());
 	ToolOffsetInverseTransform(moveBuffer.coords, currentUserPosition);
@@ -4271,7 +4284,8 @@ void GCodes::RestorePosition(const RestorePoint& rp, GCodeBuffer *gb)
 }
 
 // Convert user coordinates to head reference point coordinates, optionally allowing for X axis mapping
-// If the X axis is mapped to some other axes not including X, then the X coordinate of coordsOut will be left unchanged. So make sure it is suitably initialised before calling this.
+// If the X axis is mapped to some other axes not including X, then the X coordinate of coordsOut will be left unchanged.
+// So make sure it is suitably initialised before calling this.
 void GCodes::ToolOffsetTransform(const float coordsIn[MaxAxes], float coordsOut[MaxAxes], AxesBitmap explicitAxes)
 {
 	const Tool * const currentTool = reprap.GetCurrentTool();
diff --git a/src/GCodes/GCodes.h b/src/GCodes/GCodes.h
index ba312f2c..cf9ab462 100644
--- a/src/GCodes/GCodes.h
+++ b/src/GCodes/GCodes.h
@@ -160,6 +160,7 @@ public:
 	float GetRawExtruderTotalByDrive(size_t extruder) const;			// Get the total extrusion since start of print, for one drive
 	float GetTotalRawExtrusion() const { return rawExtruderTotal; }		// Get the total extrusion since start of print, all drives
 	float GetBabyStepOffset() const { return currentBabyStepZOffset; }	// Get the current baby stepping Z offset
+	const float *GetUserPosition() const { return currentUserPosition; }	// Return the current user position
 
 	RegularGCodeInput *GetHTTPInput() const { return httpInput; }
 	RegularGCodeInput *GetTelnetInput() const { return telnetInput; }
@@ -292,42 +293,42 @@ private:
 	bool ToolHeatersAtSetTemperatures(const Tool *tool, bool waitWhenCooling) const; // Wait for the heaters associated with the specified tool to reach their set temperatures
 	void GenerateTemperatureReport(const StringRef& reply) const;				// Store a standard-format temperature report in reply
 	OutputBuffer *GenerateJsonStatusResponse(int type, int seq, ResponseSource source) const;	// Generate a M408 response
-	void CheckReportDue(GCodeBuffer& gb, const StringRef& reply) const;		// Check whether we need to report temperatures or status
+	void CheckReportDue(GCodeBuffer& gb, const StringRef& reply) const;			// Check whether we need to report temperatures or status
 
-	void SavePosition(RestorePoint& rp, const GCodeBuffer& gb) const;	// Save position to a restore point
-	void RestorePosition(const RestorePoint& rp, GCodeBuffer *gb);		// Restore user position from a restore point
+	void SavePosition(RestorePoint& rp, const GCodeBuffer& gb) const;			// Save position to a restore point
+	void RestorePosition(const RestorePoint& rp, GCodeBuffer *gb);				// Restore user position from a restore point
 
-	void SetAllAxesNotHomed();											// Flag all axes as not homed
+	void SetAllAxesNotHomed();													// Flag all axes as not homed
 	void SetMachinePosition(const float positionNow[DRIVES], bool doBedCompensation = true); // Set the current position to be this
-	void GetCurrentUserPosition();										// Get the current position form the Move class
+	void UpdateCurrentUserPosition();											// Get the current position from the Move class
 	void ToolOffsetTransform(const float coordsIn[MaxAxes], float coordsOut[MaxAxes], AxesBitmap explicitAxes = 0);	// Convert user coordinates to head reference point coordinates
 	void ToolOffsetInverseTransform(const float coordsIn[MaxAxes], float coordsOut[MaxAxes]);	// Convert head reference point coordinates to user coordinates
-	const char *TranslateEndStopResult(EndStopHit es);					// Translate end stop result to text
-	GCodeResult RetractFilament(GCodeBuffer& gb, bool retract);			// Retract or un-retract filaments
-	GCodeResult LoadFilament(GCodeBuffer& gb, const StringRef& reply);		// Load the specified filament into a tool
-	GCodeResult UnloadFilament(GCodeBuffer& gb, const StringRef& reply);		 // Unload the current filament from a tool
+	const char *TranslateEndStopResult(EndStopHit es);							// Translate end stop result to text
+	GCodeResult RetractFilament(GCodeBuffer& gb, bool retract);					// Retract or un-retract filaments
+	GCodeResult LoadFilament(GCodeBuffer& gb, const StringRef& reply);			// Load the specified filament into a tool
+	GCodeResult UnloadFilament(GCodeBuffer& gb, const StringRef& reply);		// Unload the current filament from a tool
 	bool ChangeMicrostepping(size_t drive, unsigned int microsteps, int mode) const;	// Change microstepping on the specified drive
 	void ListTriggers(const StringRef& reply, TriggerInputsBitmap mask);		// Append a list of trigger inputs to a message
-	void CheckTriggers();												// Check for and execute triggers
-	void CheckFilament();												// Check for and respond to filament errors
-	void CheckHeaterFault();											// Check for and respond to a heater fault, returning true if we should exit
-	void DoEmergencyStop();												// Execute an emergency stop
+	void CheckTriggers();														// Check for and execute triggers
+	void CheckFilament();														// Check for and respond to filament errors
+	void CheckHeaterFault();													// Check for and respond to a heater fault, returning true if we should exit
+	void DoEmergencyStop();														// Execute an emergency stop
 
-	void DoPause(GCodeBuffer& gb, PauseReason reason, const char *msg)	// Pause the print
+	void DoPause(GCodeBuffer& gb, PauseReason reason, const char *msg)			// Pause the print
 		pre(resourceOwners[movementResource] = &gb);
 
 #if HAS_VOLTAGE_MONITOR || HAS_SMART_DRIVERS
-	bool DoEmergencyPause();											// Do an emergency pause following loss of power or a motor stall
+	bool DoEmergencyPause();													// Do an emergency pause following loss of power or a motor stall
 #endif
 
-	void SetMappedFanSpeed();											// Set the speeds of fans mapped for the current tool
-	void SaveFanSpeeds();												// Save the speeds of all fans
+	void SetMappedFanSpeed();													// Set the speeds of fans mapped for the current tool
+	void SaveFanSpeeds();														// Save the speeds of all fans
 
-	GCodeResult SetOrReportZProbe(GCodeBuffer& gb, const StringRef &reply);	// Handle M558
+	GCodeResult SetOrReportZProbe(GCodeBuffer& gb, const StringRef &reply);		// Handle M558
 	GCodeResult DefineGrid(GCodeBuffer& gb, const StringRef &reply);			// Define the probing grid, returning true if error
-	bool LoadHeightMap(GCodeBuffer& gb, const StringRef& reply) const;		// Load the height map from file
-	bool SaveHeightMap(GCodeBuffer& gb, const StringRef& reply) const;		// Save the height map to file
-	GCodeResult ProbeGrid(GCodeBuffer& gb, const StringRef& reply);			// Start probing the grid, returning true if we didn't because of an error
+	bool LoadHeightMap(GCodeBuffer& gb, const StringRef& reply) const;			// Load the height map from file
+	bool SaveHeightMap(GCodeBuffer& gb, const StringRef& reply) const;			// Save the height map to file
+	GCodeResult ProbeGrid(GCodeBuffer& gb, const StringRef& reply);				// Start probing the grid, returning true if we didn't because of an error
 	GCodeResult CheckOrConfigureTrigger(GCodeBuffer& gb, const StringRef& reply, int code);	// Handle M581 and M582
 	GCodeResult UpdateFirmware(GCodeBuffer& gb, const StringRef &reply);		// Handle M997
 	GCodeResult SendI2c(GCodeBuffer& gb, const StringRef &reply);				// Handle M260
@@ -431,8 +432,9 @@ private:
 	float arcSegmentLength;						// Length of segments that we split arc moves into
 
 #if SUPPORT_WORKPLACE_COORDINATES
+	static const size_t NumCoordinateSystems = 9;
 	unsigned int currentCoordinateSystem;
-	float workplaceCoordinates[10][MaxAxes];	// Workplace coordinate offsets
+	float workplaceCoordinates[NumCoordinateSystems][MaxAxes];	// Workplace coordinate offsets
 #else
 	float axisOffsets[MaxAxes];					// M206 axis offsets
 #endif
diff --git a/src/GCodes/GCodes2.cpp b/src/GCodes/GCodes2.cpp
index dea916c1..3be09f1f 100644
--- a/src/GCodes/GCodes2.cpp
+++ b/src/GCodes/GCodes2.cpp
@@ -89,13 +89,13 @@ bool GCodes::HandleGcode(GCodeBuffer& gb, const StringRef& reply)
 {
 	GCodeResult result = GCodeResult::ok;
 	const int code = gb.GetCommandNumber();
-	if (simulationMode != 0 && code > 4 && code != 10 && code != 11 && code != 20 && code != 21 && code != 90 && code != 91 && code != 92)
+	if (simulationMode != 0 && code > 4 && code != 10 && code != 11 && code != 20 && code != 21 && (code < 53 || code > 59) && (code < 90 || code > 92))
 	{
-		return true;			// we only simulate some gcodes
+		return true;					// we only simulate some gcodes
 	}
 	if (gb.MachineState().runningM502 && code != 31)
 	{
-		return true;			// when running M502 the only gcode we execute is G31
+		return true;					// when running M502 the only gcode we execute is G31
 	}
 
 	switch (code)
@@ -272,26 +272,35 @@ bool GCodes::HandleGcode(GCodeBuffer& gb, const StringRef& reply)
 		break;
 
 #if SUPPORT_WORKPLACE_COORDINATES
-	case 53:	// Switch to coordinate system 0
+	case 53:	// Temporarily use machine coordinates
+		gb.MachineState().useMachineCoordinates = true;
+		break;
+
 	case 54:	// Switch to coordinate system 1
 	case 55:	// Switch to coordinate system 2
 	case 56:	// Switch to coordinate system 3
 	case 57:	// Switch to coordinate system 4
 	case 58:	// Switch to coordinate system 5
 	case 59:	// Switch to coordinate system 6,7,8,9
+		if (!LockMovementAndWaitForStandstill(gb))
 		{
-			unsigned int cs = code - 53;
+			return false;
+		}
+		{
+			unsigned int cs = code - 54;
 			if (code == 59)
 			{
 				const int8_t fraction = gb.GetCommandFraction();
 				if (fraction >= 0)
 				{
-					cs += (unsigned int) fraction;
+					cs += (unsigned int)fraction;
 				}
 			}
-			if (cs < ARRAY_SIZE(workplaceCoordinates))
+			if (cs < NumCoordinateSystems)
 			{
-				currentCoordinateSystem = cs;
+				currentCoordinateSystem = cs;											// this is the zero-base coordinate system number
+				gb.MachineState().useMachineCoordinates = false;
+				ToolOffsetInverseTransform(moveBuffer.coords, currentUserPosition);		// update user coordinates
 			}
 			else
 			{
@@ -1239,8 +1248,15 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 		}
 		break;
 
-	case 104:
 	case 109: // Deprecated in RRF, but widely generated by slicers
+		if (   !LockMovementAndWaitForStandstill(gb)		// wait until movement has finished
+			|| !IsCodeQueueIdle()							// also wait until deferred command queue has caught up to avoid out-of-order execution
+		   )
+		{
+			return false;
+		}
+		// no break
+	case 104:
 		// New behaviour from 1.20beta12:
 		// M109 Snnn
 		// - If no tools are active, set Tool 0 to active
@@ -1388,6 +1404,13 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 		break;
 
 	case 116: // Wait for set temperatures
+		if (   !LockMovementAndWaitForStandstill(gb)		// wait until movement has finished
+			|| !IsCodeQueueIdle()							// also wait until deferred command queue has caught up to avoid out-of-order execution
+		   )
+		{
+			return false;
+		}
+
 		if (!cancelWait)
 		{
 			bool seen = false;
@@ -1694,12 +1717,18 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 
 	case 190: // Set bed temperature and wait
 	case 191: // Set chamber temperature and wait
+		if (   !LockMovementAndWaitForStandstill(gb)		// wait until movement has finished
+			|| !IsCodeQueueIdle()							// also wait until deferred command queue has caught up to avoid out-of-order execution
+		   )
+		{
+			return false;
+		}
 		{
 			// Check if the heater index is passed
-			int index = gb.Seen('P') ? gb.GetIValue() : 0;
-			if (index < 0 || index >= (int)((code == 190) ? NumBedHeaters : NumChamberHeaters))
+			const uint32_t index = gb.Seen('P') ? gb.GetUIValue() : 0;
+			if (index >= ((code == 190) ? NumBedHeaters : NumChamberHeaters))
 			{
-				reply.printf("Invalid heater index '%d'", index);
+				reply.printf("Invalid heater index '%" PRIu32 "'", index);
 				result = GCodeResult::error;
 				break;
 			}
@@ -2139,6 +2168,11 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 					break;
 				}
 
+				if (sParam >= 2 && !LockMovementAndWaitForStandstill(gb))
+				{
+					return false;					// if it's a blocking message, wait for movement to stop before displaying it
+				}
+
 				float tParam;
 				if (sParam == 0 || sParam == 1)
 				{
@@ -3046,7 +3080,16 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 				}
 				if (gb.Seen('T'))
 				{
-					platform.SetDriverStepTiming(drive, gb.GetFValue());
+					float timings[4];
+					size_t numTimings = ARRAY_SIZE(timings);
+					gb.GetFloatArray(timings, numTimings, true);
+					if (numTimings != ARRAY_SIZE(timings))
+					{
+						reply.copy("bad timing parameter");
+						result = GCodeResult::error;
+						break;
+					}
+					platform.SetDriverStepTiming(drive, timings);
 					seen = true;
 				}
 				bool badParameter = false;
@@ -3063,14 +3106,18 @@ bool GCodes::HandleMcode(GCodeBuffer& gb, const StringRef& reply)
 				}
 				if (badParameter)
 				{
-					platform.Message(ErrorMessage, "M569 no longer accepts XYZE parameters; use M584 instead\n");
+					reply.copy("M569 no longer accepts XYZE parameters; use M584 instead");
+					result = GCodeResult::error;
 				}
 				else if (!seen)
 				{
-					reply.printf("A %d sends drive %u forwards, a %d enables it, and the minimum pulse width is %.1f microseconds",
-								(int)platform.GetDirectionValue(drive), drive,
-								(int)platform.GetEnableValue(drive),
-								(double)platform.GetDriverStepTiming(drive));
+					float timings[4];
+					platform.GetDriverStepTiming(drive, timings);
+					reply.printf("Drive %u runs %s, active %s enable, step timing %.1f,%.1f,%.1f,%.1f microseconds",
+								drive,
+								(platform.GetDirectionValue(drive)) ? "forwards" : "in reverse",
+								(platform.GetEnableValue(drive)) ? "high" : "low",
+								(double)timings[0], (double)timings[1], (double)timings[2], (double)timings[3]);
 				}
 			}
 		}
diff --git a/src/GCodes/GCodes3.cpp b/src/GCodes/GCodes3.cpp
index 9ab7291b..1d0e20f6 100644
--- a/src/GCodes/GCodes3.cpp
+++ b/src/GCodes/GCodes3.cpp
@@ -198,7 +198,7 @@ GCodeResult GCodes::GetSetWorkplaceCoordinates(GCodeBuffer& gb, const StringRef&
 	if (gb.Seen('P'))
 	{
 		const uint32_t cs = gb.GetIValue();
-		if (cs < ARRAY_SIZE(workplaceCoordinates))
+		if (cs > 0 && cs <= NumCoordinateSystems)
 		{
 			bool seen = false;
 			for (size_t axis = 0; axis < numVisibleAxes; axis++)
@@ -214,8 +214,8 @@ GCodeResult GCodes::GetSetWorkplaceCoordinates(GCodeBuffer& gb, const StringRef&
 						}
 						seen = true;
 					}
-					workplaceCoordinates[cs][axis] = (compute)
-														? coord - currentUserPosition[axis] + workplaceCoordinates[currentCoordinateSystem][axis]
+					workplaceCoordinates[cs - 1][axis] = (compute)
+														? currentUserPosition[axis] - coord + workplaceCoordinates[currentCoordinateSystem][axis]
 															: coord;
 				}
 			}
@@ -226,10 +226,10 @@ GCodeResult GCodes::GetSetWorkplaceCoordinates(GCodeBuffer& gb, const StringRef&
 			}
 			else
 			{
-				reply.printf("Coordinates of workplace %" PRIu32 ":", cs);
+				reply.printf("Origin of workplace %" PRIu32 ":", cs);
 				for (size_t axis = 0; axis < numVisibleAxes; axis++)
 				{
-					reply.catf(" %c%.2f", axisLetters[axis], (double)workplaceCoordinates[cs][axis]);
+					reply.catf(" %c%.2f", axisLetters[axis], (double)workplaceCoordinates[cs - 1][axis]);
 				}
 			}
 			return GCodeResult::ok;
@@ -345,6 +345,7 @@ GCodeResult GCodes::SetOrReportZProbe(GCodeBuffer& gb, const StringRef &reply)
 	{
 		platform.SetZProbeType(gb.GetIValue());
 		seenType = true;
+		DoDwellTime(gb, 100);				// delay a little to allow the averaging filters to accumulate data from the new source
 	}
 
 	ZProbe params = platform.GetCurrentZProbeParameters();
diff --git a/src/Movement/DDA.cpp b/src/Movement/DDA.cpp
index a0bf5aab..45250ae9 100644
--- a/src/Movement/DDA.cpp
+++ b/src/Movement/DDA.cpp
@@ -1440,7 +1440,9 @@ pre(state == frozen)
 	return true;	// schedule another interrupt immediately
 }
 
-uint32_t DDA::maxReps = 0;		// this holds he maximum ISR loop count
+uint32_t DDA::maxReps = 0;					// this holds he maximum ISR loop count
+uint32_t DDA::lastStepLowTime = 0;
+uint32_t DDA::lastDirChangeTime = 0;
 
 // This is called by the interrupt service routine to execute steps.
 // It returns true if it needs to be called again on the DDA of the new current move, otherwise false.
@@ -1449,7 +1451,7 @@ uint32_t DDA::maxReps = 0;		// this holds he maximum ISR loop count
 bool DDA::Step()
 {
 	Platform& platform = reprap.GetPlatform();
-	uint32_t lastStepPulseTime = platform.GetInterruptClocks();
+	uint32_t lastStepPulseTime = lastStepLowTime;
 	bool repeat;
 	uint32_t numReps = 0;
 	do
@@ -1481,7 +1483,7 @@ bool DDA::Step()
 //if (t3 < minCalcTime) minCalcTime = t3;
 		}
 
-		if ((driversStepping & platform.GetSlowDrivers()) == 0)			// if not using any external drivers
+		if ((driversStepping & platform.GetSlowDriversBitmap()) == 0)			// if not using any external drivers
 		{
 			// 3. Step the drivers
 			Platform::StepDriversHigh(driversStepping);					// generate the steps
@@ -1489,22 +1491,27 @@ bool DDA::Step()
 		else
 		{
 			// 3. Step the drivers
-			while (Platform::GetInterruptClocks() - lastStepPulseTime < platform.GetSlowDriverClocks()) {}
+			uint32_t now;
+			do
+			{
+				now = Platform::GetInterruptClocks();
+			}
+			while (now - lastStepPulseTime < platform.GetSlowDriverStepLowClocks() || now - lastDirChangeTime < platform.GetSlowDriverDirSetupClocks());
 			Platform::StepDriversHigh(driversStepping);					// generate the steps
 			lastStepPulseTime = Platform::GetInterruptClocks();
 
 			// 3a. Reset all step pins low. Do this now because some external drivers don't like the direction pins being changed before the end of the step pulse.
-			while (Platform::GetInterruptClocks() - lastStepPulseTime < platform.GetSlowDriverClocks()) {}
+			while (Platform::GetInterruptClocks() - lastStepPulseTime < platform.GetSlowDriverStepHighClocks()) {}
 			Platform::StepDriversLow();									// set all step pins low
-			lastStepPulseTime = Platform::GetInterruptClocks();
+			lastStepLowTime = lastStepPulseTime = Platform::GetInterruptClocks();
 		}
 
 		// 4. Remove those drives from the list, calculate the next step times, update the direction pins where necessary,
 		//    and re-insert them so as to keep the list in step-time order.
 		//    Note that the call to CalcNextStepTime may change the state of Direction pin.
-		DriveMovement *dmToInsert = firstDM;						// head of the chain we need to re-insert
-		firstDM = dm;												// remove the chain from the list
-		while (dmToInsert != dm)									// note that both of these may be nullptr
+		DriveMovement *dmToInsert = firstDM;							// head of the chain we need to re-insert
+		firstDM = dm;													// remove the chain from the list
+		while (dmToInsert != dm)										// note that both of these may be nullptr
 		{
 			const bool hasMoreSteps = (isDeltaMovement && dmToInsert->drive < DELTA_AXES)
 					? dmToInsert->CalcNextStepTimeDelta(*this, true)
@@ -1518,7 +1525,7 @@ bool DDA::Step()
 		}
 
 		// 5. Reset all step pins low. We already did this if we are using any external drivers, but doing it again does no harm.
-		Platform::StepDriversLow();									// set all step pins low
+		Platform::StepDriversLow();										// set all step pins low
 
 		// 6. Check for move completed
 		if (firstDM == nullptr)
@@ -1542,8 +1549,8 @@ bool DDA::Step()
 		// However, following a move that checks endstops or the Z probe, we always wait for the move to complete before we schedule another, so this doesn't matter.
 		const uint32_t finishTime = moveStartTime + clocksNeeded;	// calculate how long this move should take
 		Move& move = reprap.GetMove();
-		move.CurrentMoveCompleted();							// tell Move that the current move is complete
-		return move.TryStartNextMove(finishTime);				// schedule the next move
+		move.CurrentMoveCompleted();								// tell Move that the current move is complete
+		return move.TryStartNextMove(finishTime);					// schedule the next move
 	}
 	return false;
 }
diff --git a/src/Movement/DDA.h b/src/Movement/DDA.h
index 5de52e3f..72124b30 100644
--- a/src/Movement/DDA.h
+++ b/src/Movement/DDA.h
@@ -127,7 +127,9 @@ public:
 	static int32_t loggedProbePositions[XYZ_AXES * MaxLoggedProbePositions];
 #endif
 
-	static uint32_t maxReps;
+	static uint32_t maxReps;										// maximum number of times that the ISR looped
+	static uint32_t lastStepLowTime;								// when we last completed a step pulse to a slow driver
+	static uint32_t lastDirChangeTime;								// when we last change the DIR signal to a slow driver
 
 private:
 	DriveMovement *FindDM(size_t drive) const;
diff --git a/src/Movement/Kinematics/LinearDeltaKinematics.cpp b/src/Movement/Kinematics/LinearDeltaKinematics.cpp
index 4704b191..7029deee 100644
--- a/src/Movement/Kinematics/LinearDeltaKinematics.cpp
+++ b/src/Movement/Kinematics/LinearDeltaKinematics.cpp
@@ -593,6 +593,12 @@ bool LinearDeltaKinematics::Configure(unsigned int mCode, GCodeBuffer& gb, const
 			if (gb.Seen('B'))
 			{
 				printRadius = gb.GetFValue();
+				// Set the axis limits so that DWC reports them correctly (they are not otherwise used for deltas, except Z min)
+				Platform& p = reprap.GetPlatform();
+				p.SetAxisMinimum(X_AXIS, -printRadius, false);
+				p.SetAxisMinimum(Y_AXIS, -printRadius, false);
+				p.SetAxisMaximum(X_AXIS, printRadius, false);
+				p.SetAxisMaximum(Y_AXIS, printRadius, false);
 				seen = true;
 			}
 			if (gb.Seen('X'))
@@ -609,15 +615,16 @@ bool LinearDeltaKinematics::Configure(unsigned int mCode, GCodeBuffer& gb, const
 			}
 			if (gb.Seen('Z'))
 			{
-				// Y tower position correction
+				// Z tower position correction
 				angleCorrections[DELTA_C_AXIS] = gb.GetFValue();
 				seen = true;
 			}
 
-			// The homed height must be done last, because it gets recalculated when some of the other factors are changed
 			if (gb.Seen('H'))
 			{
 				homedHeight = gb.GetFValue();
+				// Set the Z axis maximum so that DWC reports it correctly (it is not otherwise used for deltas)
+				reprap.GetPlatform().SetAxisMaximum(Z_AXIS, homedHeight, false);
 				seen = true;
 			}
 
diff --git a/src/Networking/W5500Ethernet/Wiznet/Internet/DHCP/dhcp.cpp b/src/Networking/W5500Ethernet/Wiznet/Internet/DHCP/dhcp.cpp
index 5b325645..2ecd0af3 100644
--- a/src/Networking/W5500Ethernet/Wiznet/Internet/DHCP/dhcp.cpp
+++ b/src/Networking/W5500Ethernet/Wiznet/Internet/DHCP/dhcp.cpp
@@ -938,7 +938,6 @@ void check_DHCP_leasedIP(void)
 
 	// IP conflict detection : ARP request - ARP reply
 	// Broadcasting ARP Request for check the IP conflict using UDP sendto() function
-	// TODO the following takes 620ms to time out - hanging for this long isn't very nice
 	sendto(DHCP_SOCKET, (const uint8_t *)"CHECK_IP_CONFLICT", 17, DHCP_allocated_ip, 5000);
 
 	// RCR value restore
diff --git a/src/Pins.h b/src/Pins.h
index 0b49ca74..ebe82597 100644
--- a/src/Pins.h
+++ b/src/Pins.h
@@ -44,4 +44,8 @@
 # define SUPPORT_12864_LCD					0
 #endif
 
+#ifndef USE_CACHE
+# define USE_CACHE							0
+#endif
+
 #endif // PINS_H__
diff --git a/src/Platform.cpp b/src/Platform.cpp
index aee8b7c6..43dcb2c2 100644
--- a/src/Platform.cpp
+++ b/src/Platform.cpp
@@ -66,7 +66,7 @@ extern "C" char *sbrk(int i);
 # error Missing feature definition
 #endif
 
-#if SAM4E && USE_CACHE
+#if USE_CACHE
 
 #include "sam/drivers/cmcc/cmcc.h"
 
@@ -114,7 +114,7 @@ const uint32_t fanMaxInterruptCount = 32;			// number of fan interrupts that we
 static volatile uint32_t fanLastResetTime = 0;		// time (microseconds) at which we last reset the interrupt count, accessed inside and outside ISR
 static volatile uint32_t fanInterval = 0;			// written by ISR, read outside the ISR
 
-const float minStepPulseTiming = 0.2;				// we assume that we always generate step high and low times at least this wide without special action
+const float MinStepPulseTiming = 0.2;				// we assume that we always generate step high and low times at least this wide without special action
 
 const LogicalPin Heater0LogicalPin = 0;
 const LogicalPin Fan0LogicalPin = 20;
@@ -197,7 +197,7 @@ void setup()
 #endif
 	RSTC->RSTC_MR = RSTC_MR_KEY_PASSWD | RSTC_MR_URSTEN;	// ignore any signal on the NRST pin for now so that the reset reason will show as Software
 
-#if SAM4E && USE_CACHE
+#if USE_CACHE
 	// Enable the cache
 	struct cmcc_config g_cmcc_cfg;
 	cmcc_get_config_defaults(&g_cmcc_cfg);
@@ -518,8 +518,11 @@ void Platform::Init()
 #endif
 	}
 
-	slowDriverStepPulseClocks = 0;								// no extended driver timing configured yet
-	slowDrivers = 0;											// assume no drivers need extended step pulse timing
+	for (uint32_t& entry : slowDriverStepTimingClocks)
+	{
+		entry = 0;												// reset all to zero as we have no known slow drivers yet
+	}
+	slowDriversBitmap = 0;										// assume no drivers need extended step pulse timing
 
 	for (size_t extr = 0; extr < MaxExtruders; ++extr)
 	{
@@ -2110,7 +2113,7 @@ void Platform::PrintUniqueId(MessageType mtype)
 // Return diagnostic information
 void Platform::Diagnostics(MessageType mtype)
 {
-#if SAM4E && USE_CACHE
+#if USE_CACHE
 	// Get the cache statistics before we start messing around with the cache
 	const uint32_t cacheCount = cmcc_get_monitor_cnt(CMCC);
 #endif
@@ -2308,8 +2311,7 @@ void Platform::Diagnostics(MessageType mtype)
 		Message(mtype, "not set\n");
 	}
 
-#if SAM4E && USE_CACHE
-	// Get the cache statistics before we start messing around with the cache
+#if USE_CACHE
 	MessageF(mtype, "Cache data hit count %" PRIu32 "\n", cacheCount);
 #endif
 
@@ -2491,7 +2493,8 @@ bool Platform::DiagnosticTest(GCodeBuffer& gb, const StringRef& reply, int d)
 	case (int)DiagnosticTestType::UnalignedMemoryAccess:	// do an unaligned memory access to test exception handling
 		deliberateError = true;
 		SCB->CCR |= SCB_CCR_UNALIGN_TRP_Msk;			// by default, unaligned memory accesses are allowed, so change that
-		(void)RepRap::ReadDword(reinterpret_cast<const char*>(dummy) + 1);	// call function in another module so it can't be optimised away
+		__DMB();										// make sure that instruction completes, don't allow prefetch
+		(void)*(reinterpret_cast<const volatile char*>(dummy) + 1);
 		break;
 
 	case (int)DiagnosticTestType::BusFault:
@@ -2502,9 +2505,9 @@ bool Platform::DiagnosticTest(GCodeBuffer& gb, const StringRef& reply, int d)
 		// I guess this can wait until we have the RTOS working though.
 		Message(WarningMessage, "There is no abort area on the SAME70");
 #elif SAM4E || SAM4S
-		(void)RepRap::ReadDword(reinterpret_cast<const char*>(0x20800000));
+		(void)*(reinterpret_cast<const volatile char*>(0x20800000));
 #elif SAM3XA
-		(void)RepRap::ReadDword(reinterpret_cast<const char*>(0x20200000));
+		(void)*(reinterpret_cast<const volatile char*>(0x20200000));
 #else
 # error
 #endif
@@ -2846,6 +2849,11 @@ bool Platform::WriteAxisLimits(FileStore *f, AxesBitmap axesProbed, const float
 // If drive >= DRIVES then we are setting an individual motor direction
 void Platform::SetDirection(size_t drive, bool direction)
 {
+	const bool isSlowDriver = (GetDriversBitmap(drive) & GetSlowDriversBitmap()) != 0;
+	if (isSlowDriver)
+	{
+		while (GetInterruptClocks() - DDA::lastStepLowTime < GetSlowDriverDirHoldClocks()) { }
+	}
 	const size_t numAxes = reprap.GetGCodes().GetTotalAxes();
 	if (drive < numAxes)
 	{
@@ -2862,6 +2870,10 @@ void Platform::SetDirection(size_t drive, bool direction)
 	{
 		SetDriverDirection(drive - DRIVES, direction);
 	}
+	if (isSlowDriver)
+	{
+		DDA::lastDirChangeTime = GetInterruptClocks();
+	}
 }
 
 // Enable a driver. Must not be called from an ISR, or with interrupts disabled.
@@ -3248,28 +3260,41 @@ void Platform::SetExtruderDriver(size_t extruder, uint8_t driver)
 	driveDriverBits[extruder + reprap.GetGCodes().GetTotalAxes()] = CalcDriverBitmap(driver);
 }
 
-void Platform::SetDriverStepTiming(size_t driver, float microseconds)
+void Platform::SetDriverStepTiming(size_t driver, const float microseconds[4])
 {
-	if (microseconds < minStepPulseTiming)
+	const uint32_t bitmap = CalcDriverBitmap(driver);
+	slowDriversBitmap &= ~bitmap;								// start by assuming this drive does not need extended timing
+	if (slowDriversBitmap == 0)
 	{
-		slowDrivers &= ~CalcDriverBitmap(driver);						// this drive does not need extended timing
+		for (uint32_t& entry : slowDriverStepTimingClocks)
+		{
+			entry = 0;											// reset all to zero if we have no known slow drivers
+		}
 	}
-	else
+
+	for (size_t i = 0; i < ARRAY_SIZE(slowDriverStepTimingClocks); ++i)
 	{
-		const uint32_t clocks = (uint32_t)(((float)DDA::stepClockRate * microseconds * 0.000001) + 0.99);	// convert microseconds to step clocks, rounding up
-		if (clocks > slowDriverStepPulseClocks)
+		if (microseconds[i] > MinStepPulseTiming)
 		{
-			slowDriverStepPulseClocks = clocks;
+			slowDriversBitmap |= CalcDriverBitmap(driver);		// this drive does need extended timing
+			const uint32_t clocks = (uint32_t)(((float)DDA::stepClockRate * microseconds[i] * 0.000001) + 0.99);	// convert microseconds to step clocks, rounding up
+			if (clocks > slowDriverStepTimingClocks[i])
+			{
+				slowDriverStepTimingClocks[i] = clocks;
+			}
 		}
-		slowDrivers |= CalcDriverBitmap(driver);						// this drive does need extended timing
 	}
 }
 
-float Platform::GetDriverStepTiming(size_t driver) const
+void Platform::GetDriverStepTiming(size_t driver, float microseconds[4]) const
 {
-	return ((slowDrivers & CalcDriverBitmap(driver)) != 0)
-			? (float)slowDriverStepPulseClocks * 1000000.0/(float)DDA::stepClockRate
-			: 0.0;
+	const bool isSlowDriver = ((slowDriversBitmap & CalcDriverBitmap(driver)) != 0);
+	for (size_t i = 0; i < 4; ++i)
+	{
+		microseconds[i] = (isSlowDriver)
+							? (float)slowDriverStepTimingClocks[i] * 1000000.0/(float)DDA::stepClockRate
+								: 0.0;
+	}
 }
 
 // Set or report the parameters for the specified fan
@@ -3292,7 +3317,7 @@ bool Platform::ConfigureFan(unsigned int mcode, int fanNum, GCodeBuffer& gb, con
 // Get current cooling fan speed on a scale between 0 and 1
 float Platform::GetFanValue(size_t fan) const
 {
-	return (fan < NUM_FANS) ? fans[fan].GetValue() : -1;
+	return (fan < NUM_FANS) ? fans[fan].GetConfiguredPwm() : -1;
 }
 
 // This is a bit of a compromise - old RepRaps used fan speeds in the range
@@ -3304,7 +3329,7 @@ void Platform::SetFanValue(size_t fan, float speed)
 {
 	if (fan < NUM_FANS)
 	{
-		fans[fan].SetValue(speed);
+		fans[fan].SetPwm(speed);
 	}
 }
 
@@ -3354,7 +3379,7 @@ void Platform::InitFans()
 	{
 #if defined(DUET_06_085)
 		// Fan 1 on the Duet 0.8.5 shares its control pin with heater 6. Set it full on to make sure the heater (if present) is off.
-		fans[1].SetValue(1.0);												// set it full on
+		fans[1].SetPwm(1.0);												// set it full on
 #else
 		// Set fan 1 to be thermostatic by default, monitoring all heaters except the default bed and chamber heaters
 		Fan::HeatersMonitoredBitmap bedAndChamberHeaterMask = 0;
@@ -3373,7 +3398,7 @@ void Platform::InitFans()
 			}
 		}
 		fans[1].SetHeatersMonitored(LowestNBits<Fan::HeatersMonitoredBitmap>(Heaters) & ~bedAndChamberHeaterMask);
-		fans[1].SetValue(1.0);												// set it full on
+		fans[1].SetPwm(1.0);												// set it full on
 #endif
 	}
 
diff --git a/src/Platform.h b/src/Platform.h
index 5977b2fa..253b66ec 100644
--- a/src/Platform.h
+++ b/src/Platform.h
@@ -406,8 +406,8 @@ public:
 	unsigned int GetDriverMicrostepping(size_t drive, int mode, bool& interpolation) const;
 	bool SetMicrostepping(size_t drive, int microsteps, int mode);
 	unsigned int GetMicrostepping(size_t drive, int mode, bool& interpolation) const;
-	void SetDriverStepTiming(size_t driver, float microseconds);
-	float GetDriverStepTiming(size_t driver) const;
+	void SetDriverStepTiming(size_t driver, const float microseconds[4]);
+	void GetDriverStepTiming(size_t driver, float microseconds[4]) const;
 	float DriveStepsPerUnit(size_t drive) const;
 	const float *GetDriveStepsPerUnit() const
 		{ return driveStepsPerUnit; }
@@ -455,8 +455,11 @@ public:
 		{ return driveDriverBits[drive]; }
 	static void StepDriversLow();							// set all step pins low
 	static void StepDriversHigh(uint32_t driverMap);		// set the specified step pins high
-	uint32_t GetSlowDrivers() const { return slowDrivers; }
-	uint32_t GetSlowDriverClocks() const { return slowDriverStepPulseClocks; }
+	uint32_t GetSlowDriversBitmap() const { return slowDriversBitmap; }
+	uint32_t GetSlowDriverStepHighClocks() const { return slowDriverStepTimingClocks[0]; }
+	uint32_t GetSlowDriverStepLowClocks() const { return slowDriverStepTimingClocks[1]; }
+	uint32_t GetSlowDriverDirSetupClocks() const { return slowDriverStepTimingClocks[2]; }
+	uint32_t GetSlowDriverDirHoldClocks() const { return slowDriverStepTimingClocks[3]; }
 
 #if SUPPORT_NONLINEAR_EXTRUSION
 	bool GetExtrusionCoefficients(size_t extruder, float& a, float& b, float& limit) const;
@@ -739,8 +742,8 @@ private:
 	AxisDriversConfig axisDrivers[MaxAxes];				// the driver numbers assigned to each axis
 	uint8_t extruderDrivers[MaxExtruders];				// the driver number assigned to each extruder
 	uint32_t driveDriverBits[2 * DRIVES];				// the bitmap of driver port bits for each axis or extruder, followed by the raw versions
-	uint32_t slowDriverStepPulseClocks;					// minimum high and low step pulse widths, in processor clocks
-	uint32_t slowDrivers;								// bitmap of driver port bits that need extended step pulse timing
+	uint32_t slowDriverStepTimingClocks[4];				// minimum step high, step low, dir setup and dir hold timing for slow drivers
+	uint32_t slowDriversBitmap;							// bitmap of driver port bits that need extended step pulse timing
 	float idleCurrentFactor;
 
 #if HAS_SMART_DRIVERS
@@ -1131,7 +1134,7 @@ inline float Platform::GetPressureAdvance(size_t extruder) const
 	return rslt;
 }
 
-// Function GetInterruptClocks() is quite long for these processors, so it is moved to Platform.cpp and no longer inlined
+// Function GetInterruptClocksInterruptsDisabled() is quite long for these processors, so it is moved to Platform.cpp and no longer inlined
 
 #else					// TCs are 32-bit
 
diff --git a/src/RepRap.cpp b/src/RepRap.cpp
index 5a486c2c..6d7e8008 100644
--- a/src/RepRap.cpp
+++ b/src/RepRap.cpp
@@ -607,6 +607,32 @@ OutputBuffer *RepRap::GetStatusResponse(uint8_t type, ResponseSource source)
 
 	// Coordinates
 	const size_t numVisibleAxes = gCodes->GetVisibleAxes();
+	// Homed axes
+	response->cat("\"axesHomed\":");
+	ch = '[';
+	for (size_t axis = 0; axis < numVisibleAxes; ++axis)
+	{
+		response->catf("%c%d", ch, (gCodes->GetAxisIsHomed(axis)) ? 1 : 0);
+		ch = ',';
+	}
+
+	// XYZ positions
+	// Coordinates may be NaNs or infinities, for example when delta or SCARA homing fails. We must replace any NaNs or infinities to avoid JSON parsing errors.
+	// Ideally we would report "unknown" or similar for axis positions that are not known because we haven't homed them, but that requires changes to both DWC and PanelDue.
+	// So we report 9999.9 instead.
+
+	// First the user coordinates
+	response->cat("],\"xyz\":");
+	const float * const userPos = gCodes->GetUserPosition();
+	ch = '[';
+	for (size_t axis = 0; axis < numVisibleAxes; axis++)
+	{
+		const float coord = userPos[axis];
+		response->catf("%c%.3f", ch, (double)((std::isnan(coord) || std::isinf(coord)) ? 9999.9 : coord));
+		ch = ',';
+	}
+
+	// Now the machine coordinates and the extruder coordinates
 	{
 		float liveCoordinates[DRIVES];
 #if SUPPORT_ROLAND
@@ -620,47 +646,27 @@ OutputBuffer *RepRap::GetStatusResponse(uint8_t type, ResponseSource source)
 			move->LiveCoordinates(liveCoordinates, GetCurrentXAxes(), GetCurrentYAxes());
 		}
 
-		if (currentTool != nullptr)
-		{
-			for (size_t i = 0; i < numVisibleAxes; ++i)
-			{
-				liveCoordinates[i] += currentTool->GetOffset(i);
-			}
-		}
-
-		// Homed axes
-		response->cat("\"axesHomed\":");
+		// Machine coordinates
+		response->catf("],\"machine\":");		// announce the machine position
 		ch = '[';
-		for (size_t axis = 0; axis < numVisibleAxes; ++axis)
+		for (size_t drive = 0; drive < numVisibleAxes; drive++)
 		{
-			response->catf("%c%d", ch, (gCodes->GetAxisIsHomed(axis)) ? 1 : 0);
+			response->catf("%c%.3f", ch, (double)liveCoordinates[drive]);
 			ch = ',';
 		}
 
 		// Actual and theoretical extruder positions since power up, last G92 or last M23
-		response->catf("],\"extr\":");		// announce actual extruder positions
+		response->catf("],\"extr\":");			// announce actual extruder positions
 		ch = '[';
 		for (size_t extruder = 0; extruder < GetExtrudersInUse(); extruder++)
 		{
 			response->catf("%c%.1f", ch, (double)liveCoordinates[gCodes->GetTotalAxes() + extruder]);
 			ch = ',';
 		}
-		if (ch == '[')
+		if (ch == '[')							// we may have no extruders
 		{
 			response->cat(ch);
 		}
-
-		// XYZ positions
-		// TODO ideally we would report "unknown" or similar for axis positions that are not known because we haven't homed them, but that requires changes to both DWC and PanelDue.
-		response->cat("],\"xyz\":");
-		ch = '[';
-		for (size_t axis = 0; axis < numVisibleAxes; axis++)
-		{
-			// Coordinates may be NaNs, for example when delta or SCARA homing fails. Replace any NaNs or infinities by 9999.9 to prevent JSON parsing errors.
-			const float coord = liveCoordinates[axis];
-			response->catf("%c%.3f", ch, (double)((std::isnan(coord) || std::isinf(coord)) ? 9999.9 : coord));
-			ch = ',';
-		}
 	}
 
 	// Current tool number
@@ -1341,19 +1347,26 @@ OutputBuffer *RepRap::GetLegacyStatusResponse(uint8_t type, int seq)
 	response->cat("]");
 
 	// Send XYZ positions
-	const size_t numAxes = gCodes->GetVisibleAxes();
-	float liveCoordinates[DRIVES];
-	move->LiveCoordinates(liveCoordinates, GetCurrentXAxes(), GetCurrentYAxes());
-	if (currentTool != nullptr)
+	const size_t numVisibleAxes = gCodes->GetVisibleAxes();
+
+	// First the user coordinates
+	response->catf(",\"pos\":");			// announce the user position
+	const float * const userPos = gCodes->GetUserPosition();
+	ch = '[';
+	for (size_t axis = 0; axis < numVisibleAxes; axis++)
 	{
-		for (size_t i = 0; i < numAxes; ++i)
-		{
-			liveCoordinates[i] += currentTool->GetOffset(i);
-		}
+		// Coordinates may be NaNs, for example when delta or SCARA homing fails. Replace any NaNs or infinities by 9999.9 to prevent JSON parsing errors.
+		const float coord = userPos[axis];
+		response->catf("%c%.3f", ch, (double)((std::isnan(coord) || std::isinf(coord)) ? 9999.9 : coord));
+		ch = ',';
 	}
-	response->catf(",\"pos\":");		// announce the XYZ position
+
+	// Now the machine coordinates
+	float liveCoordinates[DRIVES];
+	move->LiveCoordinates(liveCoordinates, GetCurrentXAxes(), GetCurrentYAxes());
+	response->catf("],\"machine\":");		// announce the machine position
 	ch = '[';
-	for (size_t drive = 0; drive < numAxes; drive++)
+	for (size_t drive = 0; drive < numVisibleAxes; drive++)
 	{
 		response->catf("%c%.3f", ch, (double)liveCoordinates[drive]);
 		ch = ',';
@@ -1406,7 +1419,7 @@ OutputBuffer *RepRap::GetLegacyStatusResponse(uint8_t type, int seq)
 	// Send the home state. To keep the messages short, we send 1 for homed and 0 for not homed, instead of true and false.
 	response->cat(",\"homed\":");
 	ch = '[';
-	for (size_t axis = 0; axis < numAxes; ++axis)
+	for (size_t axis = 0; axis < numVisibleAxes; ++axis)
 	{
 		response->catf("%c%d", ch, (gCodes->GetAxisIsHomed(axis)) ? 1 : 0);
 		ch = ',';
@@ -1459,7 +1472,7 @@ OutputBuffer *RepRap::GetLegacyStatusResponse(uint8_t type, int seq)
 	{
 		// Add the static fields
 		response->catf(",\"geometry\":\"%s\",\"axes\":%u,\"axisNames\":\"%s\",\"volumes\":%u,\"numTools\":%u,\"myName\":",
-						move->GetGeometryString(), numAxes, gCodes->GetAxisLetters(), NumSdCards, GetNumberOfContiguousTools());
+						move->GetGeometryString(), numVisibleAxes, gCodes->GetAxisLetters(), NumSdCards, GetNumberOfContiguousTools());
 		response->EncodeString(myName.c_str(), myName.MaxLength(), false);
 		response->cat(",\"firmwareName\":");
 		response->EncodeString(FIRMWARE_NAME, strlen(FIRMWARE_NAME), false);
@@ -1862,12 +1875,6 @@ bool RepRap::WriteToolParameters(FileStore *f) const
 	return a/b;
 }
 
-// Helper function for diagnostic tests in Platform.cpp, to cause a deliberate unaligned memory read
-/*static*/ uint32_t RepRap::ReadDword(const char* p)
-{
-	return *reinterpret_cast<const uint32_t*>(p);
-}
-
 // Report an internal error
 void RepRap::ReportInternalError(const char *file, const char *func, int line) const
 {
diff --git a/src/RepRap.h b/src/RepRap.h
index 0a8daa2f..145d23e5 100644
--- a/src/RepRap.h
+++ b/src/RepRap.h
@@ -114,7 +114,6 @@ public:
 	void ReportInternalError(const char *file, const char *func, int line) const;	// Report an internal error
 
 	static uint32_t DoDivide(uint32_t a, uint32_t b);		// helper function for diagnostic tests
-	static uint32_t ReadDword(const char* p);				// helper function for diagnostic tests
 
 private:
 	static void EncodeString(StringRef& response, const char* src, size_t spaceToLeave, bool allowControlChars = false, char prefix = 0);
diff --git a/src/Version.h b/src/Version.h
index d90862b8..13abf823 100644
--- a/src/Version.h
+++ b/src/Version.h
@@ -9,11 +9,11 @@
 #define SRC_VERSION_H_
 
 #ifndef VERSION
-# define VERSION "1.21RC4"
+# define VERSION "1.21RC5"
 #endif
 
 #ifndef DATE
-# define DATE "2018-03-11 build 1"
+# define DATE "2018-03-16 build 1"
 #endif
 
 #define AUTHORS "reprappro, dc42, chrishamm, t3p3, dnewman"