HDPE , RepRap hardware and 200 Deg C.

I am using HDPE because like Nophead is the easiest to acquire. From reading nopheads blog I have found I realy need 200 C. to get anything to stick which I have found to be true so far. I am using formica to extrude on an so far my bigest problem is I can not get my temp past 160 c. so here is my list from a profile test after 146 Deg I started to get -273 ??

Thanks Bruce W.

If you're still using the standard firmware, you may want to try switching to andreas' latest firmware. The standard firmware had a lot of problems with temperature readings getting interrupted by other things like serial communication, I think that's where you get the -273s. Andreas' is much more resistant to that sort of thing (I've never seen a completely invalid reading from it). I think the latest version is in the "losing notification-messages from the stepmotor devices" thread in the software forum.

After that, you may still have problems getting accurate readings at temperatures near 200 degrees. Some thermistor/capacitor combinations will never report more than 170 degrees or so no matter how hot you get the extruder, so keep in mind that just because the software gives you good readings for 0 - 160C, it may be way off by the time you hit 175C. Don't trust it until you've calibrated at those temperatures.

Sounds very dangerous if the temp measurement limits at 170+. Does that mean it goes into thermal runaway if you request a higher temperature?

---

[www.hydraraptor.blogspot.com]

EMF and I have been going over this very point for the last couple of weeks.
I have attached my excel file of a test calibration. The wiggle in Fluke thermocouple data at the higher temperatures is because the calibration routine thought that the temperatures had stabilized, but they really had not. EMFs data seemed to show that housekeeping overhead in the firmware was dominating the measurements for short readings (High temperatures). It is definitely not safe to trust the calibration routine above 150C. My thermocouple was reading 274C when the thermistor was in the 170s. I had hoped that a bigger cap would move the temperature window up, but not only did the window not move up the low temperature tracking diverged as well. I also tried a series resistor which I didn't really expect to work (because of the exponentials in the beta equations), but was hoping that I might be able to find a stable operating point around the temperatures of interest. The series resistor didn't work either. I have tried adjusting the preferences for the cap, beta, and power slope to get the temperature loop to close in the 200C and above range, but have not yet succeeded. I am continuing to experiment to see if there is any solution.

All help is welcome.

Best,
Dan

Edited 1 time(s). Last edit at 11/05/2007 01:33AM by Dan Putman.

The extruder heat profiler will let the temperature spiral out of control. The temperature measurement bug makes the temperature stop climbing, and that makes the heat profiler bump up to the next heat setting each minute, finally giving up when the hottest setting won't budge the temperature. Luckily, it *does* give up, so you probably won't be cooking the heater at high settings for more than 10 minutes or so before it aborts the test.

As for the rest of the software, I'm not sure if it will run away. If you ran the extruder calibrator in a range where the thermistor gives accurate results (<150), your slope/intercept will give you pretty good values for the heater settings needed at higher temperatures. I think this will let the heater climb past your desired setting since it doesn't know that it should shut off, but I don't think it will keep increasing the heater setting, so you won't go that far out of control. I haven't tested this side of things too much though.

I've discovered another benefit of having a resistor in parallel with the thermistor. Without it, if you disconnect the thermistor, the firmware stops responding to messages until you reconnect it. With a parallel resistor, you'd just get a reading equal to the parallel resistor's value and you could issue an appropriate error.

Well it seems like I have not gotten more then 170 deg. Thanks for the info on the bug well I am working on finishing up my machine I am working on end stops now and dialing in so hopfully there will be a fix in the next couple of weeks..

Bruce W.

I am also using Andreas's V7 of the firmware but with the recommended ReRap Cap, Resistor, and Thermistor..

Edited 1 time(s). Last edit at 11/04/2007 09:16PM by Bruce Wattendorf.

Dan,
It's not a series resistor you need, it's a parallel one but you will need to modify the host s/w to take it into account. I haven't got to grips with the host s/w myself yet but all the maths required is here :- [hydraraptor.blogspot.com].

As for using the slope/intercept to predict the correct heater power setting, that only works for static conditions. HDPE flow requires up to about another 3W and if you use a fan, a lot more power is required.

My system simply calculates the reading it should get from the thermistor at the desired temperature, sends that to the extruder controller and that just does on off control. I get about +/- 3C swing around the target value. I don't think it is worth any more complexity than that.

---

[www.hydraraptor.blogspot.com]

Hi Chris,

Understood. The series resistor was a vain attempt to increase the measurement time of the "resistance to time" poor man's ADC of the current electronics. As I mentioned above EMF has some data from fixed resistor measurements that suggests that there is a fixed minimum charge time for the capacitor that is dominating the measurements at low resistances. At 240C the beta equation says we are looking at somewhere around 34 ohms for the "standard thermistor"

A parallel resistor will linearize the thermistor readings which is a good thing, but it will also lower the total resistance at high temperature. It seems that it is the low resistances (short capacitor charge times) that are not being measured (reported?) accurately. That being said I will try adding a parallel resistor and see if I get usuable values.

The slope is used to compute a pair of pwm values for heater control that are paired with a pair of temperatures (I'm too cold, I'm to hot). All of these values are computed in the host software. This all works well enough in the 150C region, the problem that is killing us is that the ADC is falling apart on the short time readings.

Because this is the standard kit for RepRap, I am trying to make it work so other RepRappers that want to use HDPE don't have to rework their extruder boards. Left to my own devices I would chuck the current system for a proper ADC and a resistance bridge like yours today.

Best,
Dan

Dan,
The idea is to use the parallel resistor and then increase the cap so the time constant is the same as the current low temp readings. I think emf tried it but it did not work for some reason.

One possible problem I see is that if your hot resistance is only 34 ohms you will be trying to source 147 ma to charge the cap to 5V, which is too much to ask of a PIC. This may be why your measurements max out. Also if the measurement is done often enough it will heat the thermistor.

Perhaps the best solution with the current electronics is to use a higher resistance thermistor, plus the linearising resistor and mod the host s/w.

---

[www.hydraraptor.blogspot.com]

I wonder what Vic has done to his machine to get to temp. for PLA???

But Nophead your solution look to be the most logical..

Ah... I bet that's the effect I've been seeing. I hadn't thought about the current needed to charge the capacitor. I had been blaming it on the PIC over-reporting the amount of time measured, but to get that to fit my data I had to assume it was giving me readings about 30us too high, and I couldn't account for more than about 10us from looking at the assembly.

A new thermistor sounds like a good way to go, but I already have two nozzles with thermistors embedded in them. I'll have to try adding a series resistor to limit the current.

Trouble is the series resistor will swamp the high temp value, reducing the accuracy just where you want it.

---

[www.hydraraptor.blogspot.com]

Right... I guess I could compensate to some degree by throwing in an even bigger capacitor. What a mess.

You could add an emitter follower to provide the current. Not sure what effect the 0.7V drop would have on the calculation. Also might be in danger of heating the thermistor.

Edit:
On second thoughts it woundn't discharge unless you had a complementary one as well.

Edited 1 time(s). Last edit at 11/05/2007 04:35PM by nophead.

---

[www.hydraraptor.blogspot.com]

I just finished reading the extruder firmware again.
As it turns out an emitter follower may still work.
The way the cap is discharged is set RA7, RA6, and RA1 as outputs and then setting them low. So the cap is discharged through the thermistor, a 10k ohm resistor and a 180 ohm resistor in parallel. Adding an emitter follower driven by RA7 would still allow RA6 and RA1 to discharge the cap. The firmware simply loops Until its comparator drops below its lowest threshold, then delays another 20us for good measure.

Possibly you could change the firmware to charge through the 10K to the second lowest comparator level and then measure the time to discharge to the lowest through the thermistor. That way you would keep the voltage across it, and hence the current, low. The down side is the time would be very short so you would have to increase the cap.

---

[www.hydraraptor.blogspot.com]

The circuit that I am looking at isn't wired in a way that what you propose can work. But a 200 ohm resistor in series with the thermistor would keep the current within the specs for the pic. The longer charge time due to the resistor would need to be subtracted from the raw thermistor readings in the host software. This would properly require adding another preference value to the extruder0 prefs.
Failing that we need a thermistor that has a resistance value at 250C of no less than 200 ohms. your suggestion of a high current high side driver also has merit.

Edited 1 time(s). Last edit at 11/05/2007 10:56PM by Dan Putman.

> The circuit that I am looking at isn't wired in a way that what you propose can work.

Not sure why you say that. The circuit I looked at had one output connected via a 10K to the cap, another connected by the thermistor. Either of these ouputs can drive high, low or be left floating so you can charge or discharge through the thermistor or resistor or both. The third connection is 180R to a pin which can be connected to the comparator to monitor it and the comparator reference is programmable.

Edited 1 time(s). Last edit at 11/06/2007 04:40AM by nophead.

---

[www.hydraraptor.blogspot.com]

I have used a 30K-thermistor from rs-components (484-0161; 191 Ohm at 200C) and a 1uF capacitor and the standard ucb-board and my changed firmware v7. I could heat it up to 185C. I think it would go further but the voltage from my power supply was to less. I get 9,5V at the connection of the nichrom-wire (6Ohm) which stands for a current of 1,6A.
at this temperature the glue for fixing the nichrom-wire and thermistor was melting (I used Araldite Epoxy 2011 from rs-components). now I have found a shop in germany which sells JBWeld and got it yesterday. I will try a 50k thermistor now (318Ohm at 200C) and use a shorter nichrom.

btw, if the duration of the measurement of the thermistor is to short (< 16 cycles), the firmware should switch off the heater. I didn't testet it: so it would be nice if someone with a 10k thermistor and small capacitor could check if the heater-led is switching off.

Chris,

I just reread your post and you are of course correct. I misunderstood what you were trying to do. What you propose should work.

All in all I think specing a new thermistor for the 240C crowd is the simplest fix.

If we want to improve the circuit beyond that then I would vote for a series resistor (current limiter) and a parallel resistor (curve linearizer) and a new thermistor. This gives us the most flexibility for making the present scheme work. It also has the property that it can all be done off board in the thermistor cable.

Best,
Dan

Best,
Dan

Yes agreed the simplest solution is higher value thermistor. I was just trying to propose a solution that does not lose accuracy for people who have already fitted the 10K thermistor.

For people who have not already built it I would recommend just a thermistor with a value > 200R at 250C, a parallel resistor to make it linear, a bigger cap and a host s/w change to take the parallel resistor into account.

I think the series resistor just throws away too much accuracy. A 1% accurate time measurement would only give approx 10% thermistor resistance accuracy.

---

[www.hydraraptor.blogspot.com]

Right on all counts.

Browsing through the available parts It looks like EPCOS B57540F104F 100K NTC thermistor will do the trick ~ 234 ohms @ 250C. BOth Mouser and RS-components stock them. I just placed an order. I will report back when I have some data.

Best,
Dan

Good news,

I received and installed an EPCOS B57550G104F 100K ohm glass thermistor in to a new extruder nozzle. The data tracks my thermocouple within ~5 degrees up to 200C. I am sure it will track reasonably at 240C as well, but I did not run the cal program that high on the first run. Still some fine tuning of Beta values and power slope intercept work to do. Check out the attached spreadsheet and compare it to the 10K thermistor data I posted up thread.

Best,
Dan

Edited 2 time(s). Last edit at 11/14/2007 10:48PM by Dan Putman.

I will be ordering some next week also.. Dan let me know the settings you are using when you figure them out..

Thanks Bruce

That's great. Unless you run into any problems with it, I say we should make your thermistor the standard one, or at least put a big warning on mentions of the regular 10k thermistor. HDPE is the only plastic most of us have convenient access to, specifying a thermistor that won't work in that range is just asking for trouble.

FWIW, I messed around with a series resistor a bit. By increasing the capacitor, I was able to get a resolution of around 2 degrees when hot. Not great, but usable. If I bump the capacitor up to get in the 1 degree range, the reading times at room temperature start getting ridiculous (didn't try it though).

While gathering more data on my 100K ohm Thermistor setup, I decided to extrude some HDPE to see how the temperatures track under dynamic conditions. I had the temperature set to 240C as measured by my thermocouple afixed to the base of the heater barrel. Every thing seemed to be working fine. I had extruded ~ a foot of HDPE when I noticed that the heater barrel was doing a little dance relative to the PTFE insulator. By the time I had shut everything down the heater barrel was cocked by ~ 15 degrees. Fast forward to this morning. I started disecting the extruder. I Loosened the tensioning screws on the screw assembly, loosened the base clamp that holds the PTFE to the base extracted the PTFE, heater and nozzle cutting off the HDPE stock a couple of inches above the top of the PTFE barrel.
I removed the nozzle, and unscrewed the heater barrel from the PTFE barrel. What I observed was about a 3/8" long plug of HDPE at the top end of the heater barrel and HDPE had been forced into the threads of the heater barrel. The PTFE was slightly bell shaped at the bottom. The good news is that HDPE doesn't stick to the PTFE so it was not one solid mass as I feared. I cleaned everything up and put everything back together. I was able to thread the heater into the PTFE barrel about another .1" and everything feels tight and solid. I am considering adding a stainless steel hose clamp to the base of the PTFE barrel to keep it from deforming. For those if you starting to run at these higher temperatures PLEASE BE CAREFUL,240C HDPE will burn you in a heartbeat. I don't know yet whether This was a one off failure. Please report your experience.

Edited 1 time(s). Last edit at 11/19/2007 01:05AM by Dan Putman.

I have my extruder test rig back together.
It is happily maintaining 240C with the extruder exerciser. Thermocouple reads 243C. Note to get this behavior I have the desired temperature set at 200C.
When reading through the extruder wiki pages this behaviour is consistent with the documentation E.G. temperature set points are set at 10 and 20% above the desired temperature. Here are my current prefs. CAP 1mfd, beta is 4085, Rz 333960 intercept 28, slope 1.232. everything is working as advertised.

Best,
Dan

Edited 1 time(s). Last edit at 11/19/2007 01:07AM by Dan Putman.

I was surprised that you could get 240 by aiming for 200. I haven't seen this effect at temperatures in the 170 range where I've done most of my longer runs, but I found the page you're mentioning and that is indeed what it says.

I looked through the host code and it seems to do something different: the target temperature is set for the temperature you request, the safety temperature is set 20% above that. The power setting when you're below the requested temperature is set to 10% above what you should need, and the power setting when you're between your desired temperature and your safety temperature is set to 1% below what you need. I think what's happening is the calculated setting to maintain your desired temperature is too high, so it's just climbing until it hits the safety temperature.

I get the feeling that 20% may have been a good choice when we were planning on extruding at 80 degrees, but these seem sub-optimal for operating in the 200 degree range.

Also, be careful about overshooting, PTFE starts to break down around 260. Might want to make the host software force the safety temperature to stay a bit under that. Thanks for the warning about the hot HDPE, I'm sure I would've learned that the hard way. Probably still will..

In my opinion this s/w is madness. You can't estimate the power required with 1% and 20% accuracy. The extruder flow rate has several watts of effect and a fan near by doubles the power required.

Just turn it on when the temperature is too low and turn it off when it is too high - KISS.

Sorry for ranting but Friday afternoons are like that

---

[www.hydraraptor.blogspot.com]

EMF,

Here are the 3 data points I measured last night:
All temperatures were measured both by Thernocouple and Thermistor and agree within expected measurement error.
Desired temp: 50C > stabilised current temp of 60C
Desired temp: 100C > stabilised current temp of 120C
Desired temp: 200C > stabilised current temp of 240C

The observed behavior is that once the system is up to temperature the loop cycles between overtemp (heater/LED off) and lower_heater_limit so it is operating as a bang bang controller only with less than full power available.

Since I was not extruding during these measurements, it is possible that with the extra heat demand that the system would cycle between upper_heater_limit and lower heater limit as intended, but I will beleive it when I see it.

I view the current behaviour of the current temperature being significantly higher than the desired temperature as a bug and a Safety hazzard.

Andreas added a "reading to high" safety check to the current firmware, but it is an absolute time (timer < 10hex if I am remembering correctly) not a temperature. Since time is a function of Vref, Reference capacitor and Thermistor characteristics as well as wdt scale this check might catch a shorted thermistor, but it is unlikely to keep the temperatures in a safe range. Given that people will make mistakes having some kind of failsafe that prevents an overtemp that will breakdown the PTFE is essential.

All in all I agree with nophead, that simpler is better.

Best,
Dan