Author Topic: DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)  (Read 451880 times)

0 Members and 2 Guests are viewing this topic.

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 736
  • Country: nl
  • Please stop pushing bullshit...
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #100 on: June 08, 2015, 05:04:41 pm »
Hi Kleinstein, :-)

The  Sziklai Darlinton is not the problem, it is stable, but you have to make him stable.
Just connecting two transistor together is asking problems...
Just a small emittor resistor on the driver and direct on the base of de driver a small resistor, BINGO and its stable  :-DD
Even with the fast Audio transitors!

Q3 can be a BD139, low cost en good.

I find al this poles a little to much...

Ik like the simple design of the Harrison/HP power supply's.
I have improved the design with good results, basically it's still the Harrison Design,
but with piek current limitor,
extreme low noise reference (output noise under full load < 5uV (22Khz bandwith)
no power of / power off abberations.
Just some "tender loving care" for de Harrison design.

But i can see dat de design of prasimix has potential, and he is learning a lot :-)
I have not used any spice for "my" design, but real measurements, yes it cost a lot of time, but now I know for sure what the performance is of that power supply, real data.
Like you told, it is difficult to "spice" wiring, a lot of people do not know, how to wire a Power Supply for Low Noise, under heave load changes.
The current wiring is best to twist and the sense wires at an angle of 90 degrees to the controle circuit.

Kind regarts,
Blackdog
Necessity is not an established fact, but an interpretation.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #101 on: June 08, 2015, 06:51:00 pm »
Getting the Sziklai Darlinton unconditional stable is not simple - even in the simulation it takes some effort. There can well be trouble with capacitive loads in a certain range. Also local stability may depend on layout if this loop is very fast. The circuit with Q3 and the MOSFET is in many aspects very similar. It uses C9 and C12 for local stability - they may not be needed as local loop gain is not that large anyway.
The nice thing about such a compound output stage is, that once you have that local loop stable the rest is more predictable.

I also think there are to many caps in the circuit, and some could be left out. The one I don't like at all is C15.
The caps at the differential stage makes things look worse than they are. I would avoid having so many caps there, as the high frequency part bypassed that amplifier anyway.

The nice thing abouts simulations is, that one can fast find out which conditions are most critical. Unless one has a low frequency (e.g. 1 kHz - 10 MHz) vector network analyzer (and the right interface parts), equivalent measurements are rather slow and difficult. Today I would not design a regulator without it. It's very true that cabling is important to - so some real measurements are still needed. Wiring not only effects noise and residual hum, but also stability, especially if one goes for high performance fast regulation. With some experience one can also include some parasitic effects in the simulation to check upfront where layout is critical, and possibly test countermeasures. A fast low impedance output like a voltage regulator can react to surprisingly small parasitic inductance - so consider including the lead inductance for the power MOSFET and the inductance of the shunt.

The measured response looks surprisingly good - I had expected worse. Still the CC regulation seems to be unstable with inductive loading. Recovery of the voltage is rather slow - but better slow than having overshoot.

@ backdog:
The measured curves look impressive. However the sharp peak in the impedance might indicate that there might be instability for some specific loads (e.g. a very low impedance capacitor in the 100 µF range - so maybe not realistic).
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #102 on: June 09, 2015, 08:19:34 am »
Do you think that another current shunt monitor op amp (replacement for LTC2057) could improve the thing with CC mode or the whole CC control has to be done in different way? I hope that we can stay with low side monitoring in that case. Thanks.

Offline Bebo Connon

  • Newbie
  • !
  • Posts: 3
Re: PCBs is arrived
« Reply #103 on: June 09, 2015, 09:58:14 am »
PCB's arrived yesterday or within 2 weeks from moment of ordering. A nice supplier (printed.cz) which I'd like to recommend.



Here is the main board where is located all circuit from Sheet 1 to 4 namely input bridge rectifier and capacitors, power pre-regulator, power post-regulator (CV and CC), bias supply pre-regulator and LDOs, OCP, CC/CV indicators and isolated sync for pre-regulators. That is a functional analog part of the PSU. On the same board is also located digital part that is used to communicate with MCU board via SPI. That mean DAC, ADC and I/O expander.

Nice work, and it is so detail. mark+
For PCB supplier -printed.cz, you mentioned, How much you cost for your pcb like above?

Bebo
 

Offline daqq

  • Super Contributor
  • ***
  • Posts: 2301
  • Country: sk
    • My site
Re: PCBs is arrived
« Reply #104 on: June 09, 2015, 10:11:38 am »
PCB's arrived yesterday or within 2 weeks from moment of ordering. A nice supplier (printed.cz) which I'd like to recommend.

Here is the main board where is located all circuit from Sheet 1 to 4 namely input bridge rectifier and capacitors, power pre-regulator, power post-regulator (CV and CC), bias supply pre-regulator and LDOs, OCP, CC/CV indicators and isolated sync for pre-regulators. That is a functional analog part of the PSU. On the same board is also located digital part that is used to communicate with MCU board via SPI. That mean DAC, ADC and I/O expander.

Nice work, and it is so detail. mark+
For PCB supplier -printed.cz, you mentioned, How much you cost for your pcb like above?

Bebo
We use printed.cz as well. They have an instant qoute system which is very close to reality. See: http://www.printed.cz/eng/shop

All in all very nice people.
Believe it or not, pointy haired people do exist!
+++Divide By Cucumber Error. Please Reinstall Universe And Reboot +++
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #105 on: June 09, 2015, 02:30:39 pm »
Low side current monitoring is not a principle problem. Auto zero OPs are generally slow, so it would be very difficult to get a bandwith in the 100s of kHz at something like 80 times gain. The simple solution is to bypass the amplifier for the fast part, very similar to the way it is done for voltage regulation. This would mean having an extra R C combination from the upper end of the shunt to the CC regulation amplifier. So it might be enough to have to components air wired and still use the old PCB. To get the right values a few simulation runs would be needed.
My initial guess would be something like 220 Ohms and 220 nF for the extra RC combination, and somthing like 100 K and 10 pF for R44 and  C17. The other point is to have a capacitor or two (e.g. 470 nF film type or ceramic and 100-470 µF low ESR electrolytic) at the output. Measurements will show how much capacitance is needed to get the CC mode stable even with Inductive load.

With this extra Bypass the AZ amplifier can be slow - so the LTC2057 is no problem, just a rather expensive version in a case where a higher voltage is not needed.
 

Offline Bebo Connon

  • Newbie
  • !
  • Posts: 3
Re: PCBs is arrived
« Reply #106 on: June 10, 2015, 06:17:16 am »
PCB's arrived yesterday or within 2 weeks from moment of ordering. A nice supplier (printed.cz) which I'd like to recommend.

Here is the main board where is located all circuit from Sheet 1 to 4 namely input bridge rectifier and capacitors, power pre-regulator, power post-regulator (CV and CC), bias supply pre-regulator and LDOs, OCP, CC/CV indicators and isolated sync for pre-regulators. That is a functional analog part of the PSU. On the same board is also located digital part that is used to communicate with MCU board via SPI. That mean DAC, ADC and I/O expander.

Nice work, and it is so detail. mark+
For PCB supplier -printed.cz, you mentioned, How much you cost for your pcb like above?

Bebo
We use printed.cz as well. They have an instant qoute system which is very close to reality. See: http://www.printed.cz/eng/shop

All in all very nice people.

Thanks for reply. and I have checked the site, but I think it is too expensive than my current PCB supplier.
For my project LED lighting/motor controllers, I need the board like this: 2 layer, 10pcs, 1.6mm 85*62mm, blue solder mask,
in printed.cz , in total is nearly €128.91, I don't know if it has included the shipping costs.That's too expensive for me.
In my current PCB supplier - PCBWay( www.pcbway.com/e ), it only needs $13, even adding shipping costs,I think this would be more acceptable for me. Following is the pic of my PCB made by pcbway. at least I think it is very good for me. quite good package and quick delivery.  I basically will cooperate with pcbway unless i find a better pcb supplier. Anyway, thanks for your reply @daqq







 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #107 on: June 10, 2015, 07:06:11 am »
That's not what we hear

https://www.eevblog.com/forum/reviews/pcbway-mini-review/

and for whatever reason people sponsoring pcbway with the /e usually don't have that many posts and strangely enough they all subscribed around the same time.

There was another one but used /L at the end, so maybe the /e are all the same sales person and the /L gave up trying here.

Then again, maybe you are legit, we'll see.
 

Offline Liv

  • Regular Contributor
  • *
  • Posts: 134
  • Country: by
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #108 on: June 10, 2015, 09:17:12 am »
The one I don't like at all is C15.

Chains C15, R36; C11, R21 are necessary. When PSU voltage feedback signal is taken by a differential amplifier (with output terminals or SENSE terminals), then we should use the output of the diff. amp. for DC feedback, and for AC feedback (frequency correction). But the output of the amplifier is attenuated, the desired depth of the correction can be achieved only when the inverting operation of the error amplifier. This scheme is a classic, you can see, for example, the Agilent U8002 scheme. Take a signal for frequency correction directly from the output of the PSU is impossible, it should be taken differentially. Otherwise it will be degraded load transient due to the voltage drop across the shunt and other chains. This scheme uses two symmetrical chains, that take close to a differential signal. But it is a compromise.

This requires to have R44 and C17, to have quite some voltage feedback as well.

We have to choose between current surges and voltage surges. Typically current surges are not so terrible for load.
PSU PSL-3604 Pulse gen. PG-872 Freq. cnt. FC-510
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #109 on: June 10, 2015, 10:11:26 am »
That's not what we hear

https://www.eevblog.com/forum/reviews/pcbway-mini-review/

and for whatever reason people sponsoring pcbway with the /e usually don't have that many posts and strangely enough they all subscribed around the same time.

There was another one but used /L at the end, so maybe the /e are all the same sales person and the /L gave up trying here.

Then again, maybe you are legit, we'll see.

yes,both www.pcbway.com and www.pcbway.com/e are OK and right for pcbway. Suffix (like /e,/L) is just a recognition for their sales people.


I guess you are an employee there as well:

https://www.eevblog.com/forum/reviews/pcbway-mini-review/msg690786/#msg690786

Should I call you Erin, or eosintickleear in the forum, or "Tombs Balsam" as well in the forum?
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #110 on: June 10, 2015, 10:21:06 am »
Hah, and now you went ahead and deleted that post, cute :)
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #111 on: June 14, 2015, 12:20:21 pm »
Low side current monitoring is not a principle problem. Auto zero OPs are generally slow, so it would be very difficult to get a bandwith in the 100s of kHz at something like 80 times gain. The simple solution is to bypass the amplifier for the fast part, very similar to the way it is done for voltage regulation. This would mean having an extra R C combination from the upper end of the shunt to the CC regulation amplifier. So it might be enough to have to components air wired and still use the old PCB. To get the right values a few simulation runs would be needed.
My initial guess would be something like 220 Ohms and 220 nF for the extra RC combination, and somthing like 100 K and 10 pF for R44 and  C17. The other point is to have a capacitor or two (e.g. 470 nF film type or ceramic and 100-470 µF low ESR electrolytic) at the output. Measurements will show how much capacitance is needed to get the CC mode stable even with Inductive load.

With this extra Bypass the AZ amplifier can be slow - so the LTC2057 is no problem, just a rather expensive version in a case where a higher voltage is not needed.

Here is an upgraded model presented in post #95. This time model comes with step response to an inductive load. I decrease shunt monitor gain from 80 to 40V/V and use OP37 instead of auto-zero LTC2057. Also an additional current protection is added (Q10).
I hope that extra RC combination is added on the right place and I tested it with C12 set to 220nF, 22nF and disconnected (Zero). There is a slight decrease in current overshoot with 220nF if that can count as an valuable improvement.




Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
PSU housing ...
« Reply #112 on: June 17, 2015, 04:04:24 pm »
I spent some time trying to put completed pre-regulators into case Modushop.biz (Hi-fi 2000) Economica EP1152820 L 280 x P 200. Used heatsink is RAD-A6023/100 which I'll probably paint in black.
« Last Edit: June 17, 2015, 04:06:19 pm by prasimix »
 

Offline codeboy2k

  • Super Contributor
  • ***
  • Posts: 1836
  • Country: ca
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #113 on: June 17, 2015, 10:05:40 pm »
That's going to be a nice case when it's done! good job!

Advice: don't paint the heatsink black, that's a fallacy, and might actually degrade the heat sink performance.  Black paint is putting a continuous barrier of something non-metallic over the surface (acrylic, latex, whatever your paint is made of)... it will be like putting a blanket over your heatsink.

Black heatsinks are actually anodized, which is a process where they increase the thickness of the natural oxide layer, and color it with metallic salts (nickel acetate, cobalt acetate, etc).  This forms an outer protective layer that is also thermally conductive, unlike a basic black paint.

You want your heatsink to be able to conduct the heat away from the aluminum into the surrounding air. If  you paint it, then the heat has to cross that painted boundary, and it will be less effective. Most heat transfer is via conduction, not radiation, so painting it black is likely to lose more in its conductivity then you gain in its radiative ability, and the net effect is negative, not positive.
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #114 on: June 18, 2015, 06:48:25 am »
That's going to be a nice case when it's done! good job!

Advice: don't paint the heatsink black, that's a fallacy, and might actually degrade the heat sink performance.  Black paint is putting a continuous barrier of something non-metallic over the surface (acrylic, latex, whatever your paint is made of)... it will be like putting a blanket over your heatsink.

Black heatsinks are actually anodized, which is a process where they increase the thickness of the natural oxide layer, and color it with metallic salts (nickel acetate, cobalt acetate, etc).  This forms an outer protective layer that is also thermally conductive, unlike a basic black paint.

You want your heatsink to be able to conduct the heat away from the aluminum into the surrounding air. If  you paint it, then the heat has to cross that painted boundary, and it will be less effective. Most heat transfer is via conduction, not radiation, so painting it black is likely to lose more in its conductivity then you gain in its radiative ability, and the net effect is negative, not positive.

Thanks for advice. One thing less to do :-+.

Offline void_error

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ro
  • I can transistor...
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #115 on: June 18, 2015, 08:28:51 am »
Nice case! How much did shipping cost?
Trust me, I'm NOT an engineer.
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #116 on: June 18, 2015, 09:34:39 am »
Nice case! How much did shipping cost?

20EUR by DHL :(

EDIT: If it is ordered from Audiophonic in France shipment is 8,90EUR (30,40EUR in total incl. VAT).
« Last Edit: June 18, 2015, 11:12:54 am by prasimix »
 

Offline rikkitikkitavi

  • Contributor
  • Posts: 28
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #117 on: June 18, 2015, 09:39:34 am »
A thin layer of black paint has absoluetly no significant thermal resistanve, but adds some to the radiative thermal dissipation togethet with the convective which is by far the largest part.

So by all means, paint it black and reduce C/W somewhat but dont expect any miracles.

 

Offline void_error

  • Frequent Contributor
  • **
  • Posts: 673
  • Country: ro
  • I can transistor...
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #118 on: June 18, 2015, 02:26:53 pm »
20EUR by DHL :(
Same as what I'd get...
EDIT: If it is ordered from Audiophonic in France shipment is 8,90EUR (30,40EUR in total incl. VAT).
Apparently it'll ship for 24.50 for me :( I might have to build a case myself...
Trust me, I'm NOT an engineer.
 

Offline codeboy2k

  • Super Contributor
  • ***
  • Posts: 1836
  • Country: ca
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #119 on: June 18, 2015, 08:29:39 pm »
A thin layer of black paint has absoluetly no significant thermal resistanve, but adds some to the radiative thermal dissipation togethet with the convective which is by far the largest part.

So by all means, paint it black and reduce C/W somewhat but dont expect any miracles.

Beg to differ. Waste of time, not proven, and can only hurt more than it can benefit you.

Also see here :
https://www.eevblog.com/forum/projects/the-performance-of-a-heatsink-with-a-black-surface/msg50858/#msg50858
https://www.eevblog.com/forum/projects/the-performance-of-a-heatsink-with-a-black-surface/msg50880/#msg50880
https://www.eevblog.com/forum/projects/the-performance-of-a-heatsink-with-a-black-surface/msg50929/#msg50929

The last one is interesting and is more along my thoughts. You already have a great heat conductor, the aluminum, why paint it with something that may not actually be a 'black body' and may reduce the thermal conductivity.  Unless you know it won't degrade it, or you know the paint you choose is specifically made for painting heatsinks, why do something when you don't know its real effect?

We engineer things. We don't do anything without knowing why.
 

Offline Liv

  • Regular Contributor
  • *
  • Posts: 134
  • Country: by
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #120 on: June 19, 2015, 11:00:15 am »
Most heat transfer is via conduction, not radiation

It is not right. The books on the calculation of the heat sink can find a formula for the coefficient of heat transfer by radiation. For example, the following link:

h = e * k * (Ts^4 - Ta^4)/(Ts - Ta)

Let e = 1.0 (emissivity)
k = 5.6704^-8 W/(m^2 * K^4)
Ts = 100°C = 373K
Ta = 25°C = 298K

For the heat sink temperature of 100°C h = 1.0 * 5.6704^-8 * (373^4 - 298^4)/(373 - 298) = 8.672 W/(m^2 * K)

For the heat sink temperature of 50°C h = 6.80 W/(m^2 * K)

For the plate heat sink it is more than the heat transfer by convection. The finned heat sink fins are irradiated with each other, the "h" is slightly lower. But still "Radiation heat transfer can represent up to 30% of heat rate in natural convection heat sinks" (see the link above).

Advice: don't paint the heatsink black, that's a fallacy, and might actually degrade the heat sink performance.

It is very easy to check. On the heat sink must attach LM317 configured as current source and dissipate to it a certain amount of heat (e.g., 20 watts). Next it is necessary to attach a temperature sensor (e.g., DS18B20 or thermocouple) and measure the settled temperature. From these data, we can easily calculate the thermal resistance of the heat sink. I was doing it for the heat sink of my PSL-3604 PSU. Before painting (heat sink was anodized gray) thermal resistance was 1.325°C/W. After painting black thermal resistance was 1.1°C/W.
« Last Edit: June 19, 2015, 11:02:06 am by Liv »
PSU PSL-3604 Pulse gen. PG-872 Freq. cnt. FC-510
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #121 on: June 19, 2015, 11:18:31 am »
The paint does not have to be black - the radiation is in the IR range, and absorbtion / emission must nor be the same as in the visible range. Most organic paints are rather absorbing in the thermal IR range. The metallic aluminum can be rather well reflecting, especially in the IR range.

Its also possible to have a black paint, that is rather transparent to IR light: black "Edding" brand water proof markers are known to be rather transparent to 950 nm IR LEDs.
 

Offline Liv

  • Regular Contributor
  • *
  • Posts: 134
  • Country: by
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #122 on: June 19, 2015, 11:55:53 am »
Table from "A Heat Transfer Textbook" by John H. Lienhard IV and John H. Lienhard V:
PSU PSL-3604 Pulse gen. PG-872 Freq. cnt. FC-510
 

Offline codeboy2k

  • Super Contributor
  • ***
  • Posts: 1836
  • Country: ca
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #123 on: June 19, 2015, 09:53:24 pm »
OK, then I'll revise my thinking on painting heatsinks.   

By the way, that's a great link to the PDF of the guys Master Thesis on the analysis of the heatsink, thanks for that.  It's great to see a detailed analysis of heatsinks, and I'll read it to update myself.  My thinking on the matter seems like it might be out of date.
 

Offline prasimixTopic starter

  • Supporter
  • ****
  • Posts: 2022
  • Country: hr
    • EEZ
Re: DIY programmable dual channel bench PSU 0-50V/3A
« Reply #124 on: June 19, 2015, 10:12:49 pm »
... and I'm going to find a fine paint in coming days :). Thanks Liv once again.  :-+


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf