POWER DESIGNS PRECISION POWER SUPPLIES

[saturation]

Again, very nice data.  High precision resistors tend to have good tempcos, but not necessarily as high as you'd like.  Alas, to get it nearer to nil you'd have to mix a + and - tempco to null each other or match them so that if its used as a divider the ratio remains stable over temp.  A simple way to do it was posted here:

http://conradhoffman.com/MML%20files/1_vref_p3.jpg 

The resistance temperature coefficient is much more important than the precision of the resistor in this case. So make sure you know what those shunt tempco's are. I have looked and at low ohms and 3W it is hard to find anything with a tempco lower than 50 or 25ppm. I see those slow oscillations in CC also.  They are there in CV mode also but are very small.

I am pretty busy with work right now so I wont be participating much for the next wee

[MBY]

The resistance temperature coefficient is much more important than the precision of the resistor in this case. So make sure you know what those shunt tempco's are. I have looked and at low ohms and 3W it is hard to find anything with a tempco lower than 50 or 25ppm. I see those slow oscillations in CC also.  They are there in CV mode also but are very small.

I am pretty busy with work right now so I wont be participating much for the next week or two but I will be back.

Trend plots of oscillations when supply is set for 10uV.  That is 1.2886uV PP oscillation with spec. of 100uV PP  we are splitting hairs here but I do wonder what causes the slow oscillation.  The choppy plot may be aliasing from the low sample rate of the Trend plot.

Yes, I was a little sloppy to call them "precision shunts" without pointing it out that I actually meant the tempco. I have an assortment with shunt resistors in 1, 2, 3 and 5 watts with incredible low tempco's. I think one of them actually is something like 2 or 5 ppm, but most are 10, 25 and 50 ppm. If and when I try to use them I will give the part numbers if I can. (Edit: The best I can find in the "low tempco"-resistor bin at the moment is 25 ppm (5W and 22 Ohm, way to high). Hmm, I'm sure that I have better ones somewhere...)

May I ask what time you have on the X-axis on the trend plot? I missed that. Thank you again for a great post! I feel a little relieved that you see the oscillations, they where not just in my head!

I my girlfriend allows it, I may do some more testing in the weekend!

[robrenz]

May I ask what time you have on the X-axis on the trend plot?

The number between the min and max is the time of the last plotted point. So you can interpolate what the x axis is per division. In this case showing minutes and seconds. It will add days also when it gets that far.

[MBY]

Ok, that gives about a 30-40 second oscillation period. I shall try to verify this in the weekend.

Now I have started up another experiment to measure the duty cycle and period of the oven. A CdS is attached to the oven lamp and my Fluke 45 is logging via RS232 to a computer. The 2005 was in "cool shutdown" for a couple of days, so I going to capture both the initial warm up cycle and the "steady state cycle" in a few hours (or a day or so I expect).

Then we can superimpose the oven cycle with the output oscillations and see if they are independent or somehow connected.

It's very interesting that you noticed a oscillation even in the CV mode. What was the load? If the load was very small or none, that almost rule out the CC circuit or shunt to have anything to do with the oscillations.

Three ideas:
1) The oven cycle interacts
2) Other thermal oscillations due to unknown factor
3) Large RC-constant somewhere in the design (maybe in some discrete amplifier circuit since the ironic rule "if you build an amplifier, then its a oscillator, if it amplifies, then you tried to build a oscillator")

[MBY]

Oh! No! Nooo! Nooooooo! Magic smoke from the 2005!!!

[robrenz]

Sorry to hear that.  but it will add some diagnostic input to the thread
I have done some more testing with the Trend plot using a function gen to understand the sampling times. I have been using 100 NPLC on my testing but now realize that is a 0.573Hz sampling rate which is definetely aliasing on the waveforms I posted. When I am able I will post more info on the Trend plot Thread https://www.eevblog.com/forum/projects-designs-and-technical-stuff/trend-plot-on-fluke-884x-meters/msg114826/#msg114826

[MBY]

Ok, what now then? The crash scene investigation has only begun, but I will need some help here I think.

The oven got really, really hot and started to give off fumes. Nothing outside of the oven seems damaged except C14 mounted on the oven assembly socket. I used huge amounts of freeze spray, but the oven kept getting warm again (unit unplugged, of course), so the hot part was deep inside and since the thermal conduction is poor, the freeze spray had little effect. I pray that the inner workings is intact but you never know. Now its cool and I started to disassemble the oven.  The outer red colored tube comes off very easy, fitted with three screws. Underneath it, is two layers of two different kinds of thermal insulation. One of them is glass wool, the other I don't know, but when I unrolled it I instantly recognized it as the heater. Its a heating filament. Underneath it, is another steel tube housing the electronics and probably the thermostat.

The heater filament measured about 1.6 kOhms (cold resistance) and the filament was lying directly on the well-conducting steel tube. My initial guess is that a short circuit in the filament<>steel tube made the current increase as well as locking up the thermostat.

I noticed, when I was freeze-spraying that the assembly clicked. After a few second of not spraying, another click. And so on for a couple of times. That was probably the thermostat cooling down below threshold, then warming up again due to thermal inertia.

What I will do now:
The filament isolation is not to be trusted, so I took away the wire. I will try to run the 2005 without heater to get a feeling if other parts are working or the electronics inside is burned. Its probably hard to get inside the inner steel tube, as it seems to be riveted (small, small, rivets) to the base.

What I need help with:
Cold resistance between pin 10 and pin 9 on the oven (if someone could please do this measurement with one of the pins desoldered [not to measure the trafo wiring or other error sources]) and thermistor cold resistance [likely to be near 0 ohm of course, but I want to be sure] I would be very grateful!

I someone could measure the temperature of the oven (by demounting the outer shield and then stick a thermocouple inside the end cap of the inner tube) I would again be very, very grateful! I want to assess if the thermostat is working or not. I tried to trigger the thermostat with my heat gun, but the two complications is this: 1) I have no idea of the proper running temp, and I don't want to overheat the assembly again, and 2) I have no good idea of where the thermostat is. Presumably inside the steel tube, but maybe in the base.

Stay tuned for pictures...

Edit: "off course", not "of cause". Edit2: NO! "of course". I will never learn English, thats for sure!

[MBY]

An intriguing and maybe important find: As I was by sheer luck measuring a CdS resistor "connected" to the oven lamp for duty cycle measurement I discovered that the oven lamp probably started to shine more bright a few minutes before I was smelling the smoke. The resistance for the CdS went down from 90 k to 30 k instantly. Edit: It was warming up, so that the oven did turn on was NOT the case. No oven lamp-resistance was 250 kOhm.

That is very strange as it only can be explained by a open (or increased resistance) heater filament or a overvoltage situation. I have measured the cheapish 50W 230->115 VAC transformer and its output is normal (111 VAC, no load). The filament was not open, it measured 1.6 kOhm. The only other thing is a short in C14 in combination with a thermostat resistance significant higher than 0, or a thermostat short circuit, again with a nominal resistance of significance. The C14 measures normally (though badly burned), 9.9 nF, no short circuit and a thermostat of higher resistance in the "on" state seems implausible. 

(I will be back with pictures and schematics to make the text a little less confusing)

[MBY]

Ok, here comes the pictures!

Edit: Comments:
C14_burned.jpg: The C14 cap is mounted on the oven leads. Its surface is burned but it measures correctly after desoldering. Apparently all the heat was transfered via the leads to the cap. Strange.

Thermal_isolation_and_heater_filament.jpg: This was the stuff under the red shield, between the shield and inner steel tube. Its glass wool (I think) and some isolation around the filament. The filament is the fine wire.

Inner_tube_with_filament_tracks.jpg: This is the inner tube. You can see "tracks" of the filament that was winded around the steel tube. The filament thread is isolated with something that today only is like a powder. I hope its not asbestos!

oven_assembly_inner_tube_end_cap_removed.jpg: The inner tube has an end cap removed and not shown. This is the view of the "core" PCB with all the critical components. I sure hope its okay, otherwise the 2005 is dead!

oven_overview.jpg: A overview of the oven. The red can to the right is the outer shield.

oven_scematics.jpg: Schematic of the oven heater. Excerpt from the original schematics. As you can see, its seems drawed by a confused retard.

Oven_kicad_schematics.jpg: My KiCad redraw of the oven heater. Hopefully a bit easier to understand. I also hope that I drawn it correctly. Edit: Stupid me! The neon bulb is of course NOT in the oven, but on the front panel! Don't mind the bulb placement in the schematics (inside the blue rectangle).

2005_duty_BatteryLog_disaster.JPG: This is the measurement of the CdS (light dependent resistor). My software was originally created for battery measurement, therefore the strange name of the picture. This has nothing to do with batteries!

Power_Designs_Model_2005_Schematic.gif: As reference, the entire original schematics and component placement.

Please help! Any input is appreciated!  My only idea so far is to run the 2005 "cold" to check if the electronics work, but I'm reluctant to plug it in!

[MBY]

Ok, I could not restrain myself from running the unit without heater. I have put one thermocouple inside the "core" of the oven assembly and one in the mains transformer, just to measure if something seems to be overheating.

So far so good, the dial is still on 9.993 V and output is 10.005 V, but of course with "huge" drift.

So, I think the electronics has survived! Yeah!!!!!

But something that is bleeding obvious is another oscillation! The freaking oven lamp is pulsating in a sawtooth manner in a period of 3 seconds! A notable dim down, then slowly rising in brightness, and then a sudden dim down again! I think I'm on to something. You see, a neon bulb is actually a active component. Sort of. Its possible to create multivibrators, logic gates and flip flops using nothing but neon bulbs, resistors and caps. Its an old dying art and I'm far to young to be an expert, but I know it could be done and I have done experiments myself. A "bulbe astabile multivibrator" is actually pretty trivial.

So, here what I think: The bulbs is creating the oscillations for some reason. There may be unintended bulbe oscillators! It is however very strange since all bulbs except the "+10V" one is driven by AC. You should have to have DC to make oscillators. What the heck is going on?

As I have no ovenized 2005 right now, but a crippled one I cannot confirm the drift-oscillations at the moment as the temp drift masks things out, but if someone is to temporary cut a wire to one or all bulbs I bet that the drift improves!

Here is a whole book on neon lamp circuits:
http://computer-refuge.org/classiccmp/neon_lamp_logic/

Edit: Another non obvious component is RT1, a disk thermistor mounted on the main PCB, well outside of the oven. NTC? PTC? The BOM doesn't say anything but 1000 ohms at 25 C. Its mounted as if it where to monitor Q4, but why? I have to spend the whole freaking night looking for "relaxation oscillators" in the vintage style, non obvious schematics...

Edit: English...

Edit: Hmm, the visible oven lamp oscillations has stopped! I dared to freeze-spray the bulb, the inner tube of the not-so-warm oven and on some random places, but the oscillations did not re-appear. Spooky shit!

And... As I was typing, the oscillations re-appeared when I toggled the ON switch! But they are much fainter now. What is this shit? A psychological phenomena? I'm seeing stuff? Confirmation bias? I getting mad! I don't even know if I should fix this with a solder iron or a shrink!

[SeanB]

It could be C14 has become conductive, and is acting as a varistor. Best tested by replacement, as it is a very elderly ceramic, and they are known to grow whiskers internally like any other ceramic capacitor.

[MBY]

I tested C14 with my UT 511 isolation tester. I tested up to 500 VDC in both directions and found no problem. As I said before I also tested it for capacitance and measured 9.9 nF, so it seems to be okay (but I will replace it anyway).

And an intermittent short would only force the oven on, not increase brightness on the oven lamp and badly overheat the oven in such short time, if not the thermostat has a significant high resistance. I found the latter to be implausible as it heat up itself.

[MBY]

I still don't know what went wrong. But as the electronics seems to work fine, only the oven needs replacement. I have started to design a new heater, but I have no good numbers on the original one.

Using a power meter in the wall socket before, I noticed that the oven consumes about 10 watt in power. But there is likely to be losses from the 230->115 VAC converter included. I also have no good idea of the ballpark temperature is to be, eg when the thermostat trigger. It should be fairly easy to measure the heater filament resistance and the regulated temperature on a functionally sound unit without the need to take it apart in atoms. The outer red shield comes off pretty easy and the end cap of the inner steel tube is also easy to remove. To measure filament resistance, one leg of the oven needs to be desoldered.

If someone where to measure the temperature and/or the filament resistance on their unit I will be very grateful! Especially the temperature would be very useful for me. My goal is to regulate at a slightly lower temperature not to trig the thermostat if it perhaps works. A lower temperature controlled by PID and a recalibration should make the unit almost new!

I'm starting a new thread on the oven project, stay tuned...

Edit: Link to the project thread: https://www.eevblog.com/forum/projects-designs-and-technical-stuff/a-new-improved-oven-for-the-pd-2005-precision-power-supply/

[andyb]

Just wanted to let you guys know that this thread compelled me to buy a PD 2005 and register for this forum...

[saturation]

Welcome, when you've a moment, give a review for us.  Since they are all used, the end-user quality can vary, but so far its been fairly tough.

Just wanted to let you guys know that this thread compelled me to buy a PD 2005 and register for this forum...

[MBY]

I'm doing some progress with my new heater. The PID regulates well, but is (I imagine) too slow when put in real service. A bigger temp change takes some time to adjust and as it happens, the electronics inside the heater (the reference amp stuff) contribute to the heat. So, when the PSU is off but the oven is on I imagine that turn on will rise the temperature, adding time to stabilize. That somewhat defeats the purpose of having the oven run on all time. I'm sure I can figure it out somehow.

Anyway, I plead to you who owns a PD2005 to help me! I really could use two key facts.
1) What is the running temp inside the oven with the thermostat drive?
2) What is the power of the heater?

These two key facts is really easy to measure. The first one only need a screwdriver.

1) The outer (red) shield is held in place by three screws in the socket. They don't need to be removed, only loosened. Then, the shield will come off if you turn it slightly and then it just separates. Inside, you will find a inner tube covered with isolation, but with a end cap that is not covered. The end cap comes off with two screws. Then you can stick a common thermocouple or whatever inside the electronics package ant measure the temperature. I'm not proposing anything destructive here! You can reassemble the end cap and the shield as easy as it comes off! I'm not suggesting that you should break or harm your 2005 for me!

2) The second does even not need dismantling anything besides removing the cover of the unit! Do like this:
A) Turn off the device (front panel switch. This is to break the transformer from the heater).
B) Disconnect the unit from mains (not only for safety, but also because you otherwise will try to measure ohms of a live wire).
C) Measure resistance between pin 9 and pin 10 on the heater assembly socket (clearly marked on the socket)
D) Measure resistance of the thermostat (in "on" position presumably), between pin 11 and 9.
There are some caps in parallel with the heater and thermostat so please measure "common" ohms on a simple multimeter, not a real LCR meter doing AC measurements. 

Its easy! A temperature and two resistances. It will not break your unit (if you aren't incredible clumsy)!

First one that comes up with those digits will be mentioned on the PCB of my second spin of my new PID regulator!
...make that the two first. One brave first, and one for confirmation!

[robrenz]

Pin 9-10  674.3 Ohm
Pin 9-11  .00839 Ohm
Oven case temperature 48.0 Deg.C

I did not want to open my oven and since the absolute temperature is not critical I feel the temp of the top of the oven with a thermocouple is sufficient. It was measured with the outer case on so it should be very close to the inner temp anyhow.

robrenz

[MBY]

Thank you robrenz! 674 ohms is much lesser than I expected. That translates to over 19 watts of heating power. Did you measure with the unit unplugged and in off-mode (otherwise you will measure the transformers primary in parallel)?

I think it is a difference between oven shield and inner temperatures. The insulation was wired around the inner heater and on the top, so isolated in all directions. I think the best spot for external temp measurements is perhaps the base as there is no room for isolation (an air or plastic gap maybe).

As you said, the absolute temperature is not critical, its only a matter of calibration. I have done the first tests of the oven controller heating the real electronics. My setpoint is 60 degrees. I shall do experiments with insulation as I don't plan to re-use the old isolation (some wool). I actually did experiment with toilet paper but I think I go with something else. I'm experimenting with a shield in alu foil and maybe I get a chepo vacuum flask.

[robrenz]

Thank you robrenz! 674 ohms is much lesser than I expected. That translates to over 19 watts of heating power. Did you measure with the unit unplugged and in off-mode (otherwise you will measure the transformers primary in parallel)?

Yes unplugged and off.

[SeanB]

If you keep under 50C then polystyrene will work as an insulator quite well.

[MBY]

Thank you robrenz! 674 ohms is much lesser than I expected. That translates to over 19 watts of heating power. Did you measure with the unit unplugged and in off-mode (otherwise you will measure the transformers primary in parallel)?

Yes unplugged and off.

Hmm, that may explain why the heating up period of my PID is so slow as I only putting in about 5.5 watts to the heater. I used Kanthal wire of 178 ohms/m (six sub-metre threads in parallel making 21 ohms totally). My setpoint is 60C, and I think I will settle there as I suspect that the original temperature is higher. Lower temp translates to longer working life and lesser power consumption (in the heater, not electronics of course) but may be more sensitive to ambient temp drift.

SeanB: Thank you for the tip. I may settle at so low as 50C, but if I find a good insulation for higher temp I will try to use a higher setpoint.

[SeanB]

The original would be a simple glass wool, a rather nasty thing to work with. Causes skin problems, and the fibres are known to be carcinogens. Try using some polyester wool as a replacement, the common wall insulating ones will work well up to about 50C with little problems. Not good if you are running at more though, it likely will melt.

[MBY]

Pin 9-10  674.3 Ohm
Pin 9-11  .00839 Ohm
Oven case temperature 48.0 Deg.C

I'm doing a second spin of the board now (pics and description in the other thread for the project). I promised to add something (copper layer, as this is home-etched) to thank you for your help. Some ideas?

The original would be a simple glass wool, a rather nasty thing to work with. Causes skin problems, and the fibres are known to be carcinogens. Try using some polyester wool as a replacement, the common wall insulating ones will work well up to about 50C with little problems. Not good if you are running at more though, it likely will melt.

Yes, it's was glass wool outside the filament and some pulverized stuff that has the same grey-blue color as asbestos just around/under the filament. A got rid of it all and is currently experimenting with insulations. Toilet paper, PU foam, cotton pads and even a vacuum flask are among those things I'm testing with. I'm actually going to fit some stuff outside of the outer shield as well. It's gonna be overkill!

[Aldobrandi]

So after thoroughly reading this thread and lusting after the stability of these supplies I decided to get myself a 2020B for my hobby work (got it for $80 plus shipping). After some very basic testing the unit seems to be in good working order and its insides are pristine looking (ill try to upload pics later). Now it was last calibrated in 1999 and i lack the equipment, let alone the skill, to properly calibrate it. I contacted a calibration lab here in NJ and they told me they'd charge me $75. Does that sound fair? Also what kind of tests can you recommend I do with my pathetic equipment (4000 counts Chinese DMM) and no oscilloscope (yet!) before I send it?

Thanks a lot in advance and a huge thanks for bringing these puppies to my attention!

[T4P]

Roughly check the voltage drift from set after letting it stabilize for half an hour, then probe the outputs for any noise if exist means caps expired