DUET bed heater, spikes and other related issues.

Hi All,

This may have been obvious to some of you, but thought I would point out this relatively simple add-on circuit.

Concern has been expressed about the heater power/current heater switching spikes and the power supply arrangements.
Looking at the Duet circuit there is a 2 pin external FET connection, so what use is that, well see the attached image.



If the bed heater ribbon is removed this can be directly plugged into this add-on board. All that is then required is to add power and this can come from some spare connectors on the supply cable.
Connection back to the DUET are simply a 2 wire thermistor lead and a 2 wire Ext_FET lead.

Further improvements (not shown to avoid confusion) are to add a catch diode across the heater or a D and R,C type circuit.
To prevent high current going down the Ext_Fet GND lead in the event of a power supply GND break, a classic sacrificial low wattage resistor or fuse can be placed in series with this connector.
I'm sure there are more add-ons such as FET switching speed mods and PSU capacitors that can also be added, but this PCB would isolate the power switching from the DUET board and is quite a simple board (ignore my track layers it was just done quickly to illustrate the point!).

---

Ormerod #007 (shaken but not stirred!)

Hi Treth,

this is intriguing - could you show the schematic too please?

I should hook up my scope or get better at reading C++ and check the firmware, but as a shortcut; do you (or anyone else reading) know if the FETs are used in PWM or just slow on/off ? Looking at PSU options in other threads (maybe too many to mention, but summarised as "ATX 12V sucks and the current rating of many versions for continuous high current draws is rubbish and we should buy much better specced regulated DC 12V supplies, or consider using mains power direct"), I'd be quite happy to use a SELV (eg 12V halogen lighting) transformer - these are cheap, and readily available from ebay, Maplins and screwfix, are rated for continuous output of up to and above [email protected] and are easily safely wired by non-electricians. The main drawback is that the output is AC and would require back to back FETs driven by an optoisolator (or an SCR based system I guess) if they needed high frequency switching. If the output is on for a few seconds, off for a few seconds, then maybe the on-board FETs could switch a relay that would switch the transformer (preferably on the output for safety of hook-up) - the extruder and bed heaters wouldn't mind AC current (but the bed LED might)...

Ray

The bed heater is on-off, the extruder heater is PWM. IMO the main design mistake is not putting any decoupling capacitors on the 12V rail local to the bed heater circuit, coupled with turning off the mosfet very fast. This results in a very short positive spike on the +12V rail. It's not too difficult for those with electronics experience to fix this, and I'm satisfied that it's no longer a problem since I modified my board, and perhaps never was. Yes the 12V line does sag a little when the bed heater comes on, but I don't think this really matters unless you need to heat the bed to a higher temperature for ABS. I've bought a 12V 300W power supply (£21-80 via eBay) so that I have the option of replacing the ATX PSU, but I'm not using it yet..

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

well you can get this from screwfix tomorrow for <£25 or this from ebay for <£5 including postage if you like to wait. and aren't worried about replacing it
Both 200W, both rated for continuous duty, and both would relieve the ATX of its most demanding task, which is heating the bed...

Ray

Ray

I am not sure that these light transformers are suitable. The Ebay one has AC output and the Screwfix unspecified so rectification and smoothing may be needed. May be easier to use a second ATX.

Rory

Now see you cover this in earlier post.

Edited 1 time(s). Last edit at 01/10/2014 07:24PM by Rory166.

Quote
rory166
Now see you cover this in earlier post.

indeed Rory, I'm trying to be cheap and lateral on the power supplies if there's no need for the heater to be run from a regulated dc psu, then why (potentially) overload the one we've got or buy a (potentialy) more epxensive one, if we can get cheap unregulated AC ones that will provide the power needed happily and cheaply (answer probably lies in demands on switching, which may not be too much of a problem if it's not high speed and if these "transformers" don't mind having their outputs switched rather than their inputs - they're too light in weight to be iron-cored copper hence my inverted commas)

Edited 1 time(s). Last edit at 01/10/2014 07:34PM by rayhicks.

Hi,

Please don't use that type of light transformer, its not precisely regulated because it designed for constant load, not better than normal ATX PSU.

Ew

yes Ew,

hence my question about whether the heaters need to be PWMed or switched on and off slowly - switching the heated bed on for say ten seconds (of constant load) then off for two seconds (of zero load), then on again for ten seconds is pretty constant for a while, no? How precisely would it need to be regulated for heating?
Halogen bulbs and the heating bed surely have similar needs - both start up with higher current demand which diminishes as they get to temperature (because of their ptc) both are partly inductive, neither need "precise" voltage regulation - what would be damaged? My mother used to tell me not to play with the light switches or I'd blow the bulbs - and she was right, they've got a pretty low thermal inertia and the stress that thermal cycling puts on white hot filaments when heated and cooled repeatedly does shorten their lives (when cycled in periods of the order of seconds particularly), but embedded heating elements don't suffer in the same way, since they only cool by a few degrees per minute.

I can see a problem with switching these transformers electronically (in terms of circuit complexity) because of the AC output, but if the cycle time is in the order of seconds, a pretty cheap relay should be able to cope for a few years, and the on-board FETs should be able to drive such a cheap relay.

Ideally (for the wiring safety reasons you mentioned in another thread when discussing encapsulated PSU's) the output of such a relay would be switched while the mains side would be wired to a plug, if the transformers can't cope with output switching, then that could be a problem of course...

Ray

Hi Ray,

Yes,I agree with you about advantage of PWM for switching the heated bed . I did suggest about PWM for heated bed in others thread too.
Info. from this forum , the duet board design seems a little sensitive on noise immunity , on-off spike will interfere or freeze/ hang the boards operation. Sometimes designing PCB need to be very careful to place ground plane layout , digital grounds are different compared to analog device or high current device, board like Duet need to have separate ground for digital circuit and analog / high current ground on PCB so that high current flow on the board will not interfere the digital circuit. Thats just my wild guess on Duet board design because I'm still waiting for my Ormerod to arrive. Can't confirm yet until I receive mine and check on oscilloscope.


Ew

Quote
rayhicks
Hi Treth,
this is intriguing - could you show the schematic too please?
Ray

Hi Ray, I just quickly produced everything in the attached pdf which will hopefully clarify the proposal.
It primarily replicates what is on the DUET board so there are no design changes. The new PCB would allow other options to be added such as spike suppression and PSU capacitors, isolation etc.,

Look forward to comments.
ExternalBedHeaterPCB.pdf

---

Ormerod #007 (shaken but not stirred!)

Great -thanks Treth

Ray

Personally, I think adding external stuff for the bed heater is overkill. I'm still using the Duet board (albeit with some mods, which may or may not be necessary) and the Alpine ATX power supply, and it has been working fine. I had one print hang before I modded the Duet board, and none since, apart from one that was caused by the attached PC going to sleep (the print restarted when I woke the PC). When I want to print with ABS, I will probably replace the ATX PSU by the 12V 300W PSU I bought via eBay to get some additional voltage; but for printing with PLA, the existing setup works just fine for me.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

@DC42, yes could be an overkill, but I'm considering this from several aspects.

1. Many people do not like to mod their DUET board, what if they cause other issues.
The add on board allows the spike suppression to be easily incorporated (I'm still looking for original post but think the R and C can be added where my add on board allows for the two "R's" space) in the FET drive, accross the load etc., it also allows for separate power leads/PSU for the bed heater the bed and additional local decoupling capacitance. This is all without mofification to the DUET PCB.

2. With my "what happens if" hat on, there is a lot of power in this part of the circuit. The DUET is a four layer board and a failure of something in this part of the circuit could damage the DUET beyond repair.

3. It could improve the DUET performance or reliability in other areas, by removing the on board power dips and any residual switching spikes. Yes the spikes may have a mod option to be eliminated but I have held back from inserting the R in the FET drive and don't like just to slug it with a C. I do support this as a preferred way of handling the switching spikes as catch diodes etc., are not curing the cause, but handling the effect.

It is up to individuals to decide whether they pursue this, so I thought I would point out this option.

Incidently, for those not concerned with the appearance it can be 'air wired' with a pcb/wirewrap type ribbon connector with the heater connections feeding from each end. The thermistor can be taken off from the middle contacts. This only leaves the FET and drive wires (with protection resistor) to add. Not recommended as a solution but it is an option!

---

Ormerod #007 (shaken but not stirred!)

Having a separate board is not a bad idea. I built a board with an opto-isolated MOSFET to switch a heatbed, I use it if I need to run different voltages, or I'm not confident about the main board and want to avoid the chance of toasting it.

You can also get SSRs for a few pounds, which are a simpler alternative and quite effective apparently.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license