HP HSTNS-PD43 PSU Hack & Feedback Circuit (Partial)

[MrEcho]

I just updated my reply with photos of the finished PSU and changed some of the content. It was actually the negative traces i drilled through.

[zheka64]

Hi.

I finally bought myself this modified PSU and I think I figured out how it's made on this board with jumpers.
It's a feedback circuit, so it won't work with just resistors. As far as I understand, it takes +12V from two points, one point to the shunt, the other after. It goes through the current sensing amplifier INA180A4 and feeds the output to the non-inverting input of the operational amplifier LM321. +5V is applied to the inverting input through a voltage divider (the value is set by a jumper). The output through a diode is fed to the same place where the voltage is set. So, the higher the current, the higher the voltage at the output of the LM321, and accordingly, the voltage and current at the output of the PSU will decrease, and vice versa.

Schematic of first PCB:


The function of the other board on the large capacitor is still not entirely clear. It has a 4-channel operational amplifier LM324 and an optocoupler for power control, with wires from it going to the power-on pins. This board might also be responsible for reducing the fan speed. Additionally, the device turns on and off in an unusual way, as if it's going through a protection trigger (there's an audible beep).

Regards.

[Xballe]

Hi All
First a question: Did you settle on a final value of resistance (10.50k ~ 10.60k) for the OVP ?
If I understand it correct, the OVP resistance (10.50k ~ 10.60k) should work for all voltages ( 12.66V - 100.1k~103.5k    to    15.50V - 10.68k ) right ?

Next: I'm in the process of building a 24 Volt power supply, two HSTNS-PD43 in series. Therefore one of them needs to be floated (low voltage ground disconnected from cage ground)
Its pretty easy to float the HSTNS-PD43, you don't even have to open the PSU to do it (as I did, to find out how simple it is)

First you have to bend out the little tab (red circle) 45 deg. should do it, then remove the screw (green circle) from the bottom, closest to the tab you just bent out.
Now the PCB should be able to move a little up and down at that end.







Now you need a M3 plastic/nylon bolt (approx. 1 – 1,5mm longer than original screw) and an isolating washer, I used 0,4mm mylar for the job. It needs to be a thin and sturdy, isolating material.
You need a tight fitting hole around the M3 bolt, due to a small protrusion around the hole on the standoff on the PCB.




The most tricky part is to get the washer in place between the case and PCB, but when its in position put the nylon bolt in and tighten, DONE !
Your PSU is now floated and can be used as the second, third.... or how many you want to stack in your final setup.

Now, a little insight as to why the washer has to be tight fitting and sturdy.
First when I took the PSU apart I expected I needed to isolate the PCB in 3 places (red rings in pic) but actually the PCB is only grounded in one spot (yellow rings)




The little protrusion I mentioned earlier is only 0,7mm high and 4mm in diameter with a M3 hole in the middle, that leaves us with a 0,5mm wide protrusion that the isolating washer has to "ride on" and stay on, hence the tight fit mentioned.







Hope this makes it a little easier if you have to float your HSTNS-PD43 

[Trurl]

Very cool Xballe! Thanks for sharing the detailed photos and pointers!

For now I'm leaning towards 10.50k resistance for the OVP (I've yet to test the PSU at high amperage*). But I would not be surprised if the resistance may need to be lowered a bit for those intending to use the PSU mostly for very high loads(say 50~100A) & voltage settings(say 15.0~15.5V) to keep the OVP from kicking in. Note that I mentioned in the OVP triggering test results post that for higher amp loads the OVP in some cases triggered even though at low amp loads it did not (under the same resistance & voltage setting).
*I'm planning on setting up a dummy load using relatively heavy gauge nichrome wire for testing at high amps. But I recently got way distracted when I found on the street a very nice Samsung fridge compressor(from a drinks refrigerator) and an empty mint condition refrigerant tank(pink for R410A = high pressure rating) among throw away items from a wedding hall business that dropped dead (Note: I'm sure you can imagine, this sort of find of "precious" oddball items(with no dings/rust or anything), in one location just hangin' out on the street, is incredibly rare, so I had to grab them... BAM!). Hence, I've been distracted into "DIY silent compressor / vacuum pump in one" design mode (fun times).  Cheers~

[stanislav]

Hi Zheka,

I bought a similar unit back in July and looked inside to understand how I can modify them myself. The seller told me that the new mod is coming where voltage and current can be regulated so my version is at fixed 14.6V and 25/50/75/100A jumpers.

So here's a circuit of the added pcba in my version. Two modification that I've made:
- added a multiplexer instead of the jumpers to manage current programmatically but the idea is that the jumpers regulate voltage coming into the op amp
- In the original mod a resistor of 4.3k was replaced with R1 and R5 on my circuit (the multiplexer adds another 300Ohm so I wanted it to be roughly the same)

Connections:
- SIG1 - A9
- SIG2 - upper node of 05C resistor at B9
- SIG3 - D5 (MC3-1)

This is from my memory but I can double check it if anyone needs it.

Some observations
1. No need to tune the potentiometer, the hack provides around 14.6V
2. There are different revisions of the unit:
- 0B(08F) - what I've gotten from the seller
- 0A(03F) - an older version with no hole for SIG3 but it's still possible to hack it by using U3A op amp output

So far the mod works well for me and I've managed to tune a PSU to get the desired output.

Here's the challenge that I don't know how to address and maybe someone can help me here. When PSU is connected to a battery, the circuit still works and consumes around 1.3W with fans barely running. Any ideas what could be done in order to prevent power consumption? Note that PSKILL, PSON pins are disconnected from GND. Appreciate any help.

Also, guys, if you have a newer version, could you check if PSU with turned off PSON/KILL pins and no AC is still running when voltage presents?

P.S. Appreciate all the above information. It's probably the only place with details for those units.

[zheka64]

Hi Stanislav,

Thanks for sharing your version of the attached PCB schematic. Mine PSU revision is 0B(10F), but it's made as I described earlier. It's likely they decided to unify the schematic so that it's the same for both revisions.

Regarding the fan working when the device is turned off and battery consumption - it's still the same. By the way, do you have a second board on the large capacitor? And do you have any thoughts on what it does besides power on control?

Regards.

[stanislav]

Thank you for checking how the fan works!

In my version there's no second pcba and I'm only guessing here. As one of two units I bought didn't work, the seller "suggested" that I've burnt it via a wrong polarity connection that wasn't the case. Turned out it was an original manufacturer problem. A week later the seller mentioned that these PSU also die from low AC voltage, something like 200-205V so given it's on the power side, it could be related to that.

[AtlanticSurfer]

Hi guys!  My aim is to get a PSU for automotive use and I’ve managed to perform the voltage mod on one of these PD43s.  Next up was to make an enclosure for it and was hoping to be able to use one those cheap Aliexpress VA meters.  I was hoping to avoid the negatives of including a big shunt and had read that the common slot supplies had an analogue current pin so I set about probing the PD43 to see if I could one.  I’ll attach the results plot below.  I did find a pin where the voltage between it and ground increased proportionally (to an extent) with load current but I’ a bit baffled that there’s a small negative voltage at zero load.  I had hoped to to use a voltage divider to emulate a 75mA shunt but I can’t think how to do that if there is a negative component.

Has anybody managed to get an analogue current measurement from the PD43 or that can suggest what I’m doing wrong?  I did double check that no load voltage and measured -0.089v instead of the original value of -0.078v.

[Xballe]

https://www.aliexpress.com/item/1005005594307828.html?spm=a2g0o.order_detail.order_detail_item.4.1b25f19cQAHSvU

I ordered one of these from Ali to try and avoid the shunt as well. Haven't seen it yet so can't comment on how it works or precision

[AtlanticSurfer]

Thanks Xballe, that certainly looks like an option.  I kept looking past any ammeters with hall sensors as I associated them (wrongly) with AC current.  I'm not concerned about precision really, just want to have an indication but I am concerned about expense and that sensor is more expensive than the PSU was.  I wonder if it's possible (perhaps using an alkaline cell and a trim pot) to lift the shunt voltage above ground?

I s your intended use also an automotive PSU?

[Trurl]

... I'm not concerned about precision really, just want to have an indication but I am concerned about expense and that sensor is more expensive than the PSU was.  I wonder if it's possible (perhaps using an alkaline cell and a trim pot) to lift the shunt voltage above ground?

I s your intended use also an automotive PSU?

You may want to consider a panel voltmeter with an ICL7107 chip as it has a "differential input" that can "zero" out the non-zero current sense voltage in unloaded state, and convert this voltmeter into an "ammeter".

In my case, I get like +26mV@0A offset voltage from the current sense voltage(CSV) - from that same first pin on the underside edge. The ICL7107 based voltmeter's input will make this offset voltage zero at its differential input. And as the PD43's current sense output voltage is about 25mV/A you'd need to set the voltage dividers along the input side accordingly to accommodate this. If interested, you can read about my experiences figuring this out in the following threads (some may not directly be related to your specific interests but some potential pitfalls along the way could be similar and useful to know about in advance):

1. ICL7107 - Is VRef Greater than 1V Possible?
https://www.eevblog.com/forum/projects/icl7107-is-vref-greater-than-1v-possible/

2. How To Make Non-Zero Initial Current Sense Voltage Zero?
https://www.eevblog.com/forum/beginners/how-to-make-non-zero-initial-current-sense-voltage-zero/

3. ICL7107 Volt Meter Hacked To An Ammeter With 2.5V Ref And In Lo Offset
* For some reason(after I edited the thread's title) clicking the link for this thread below is not working (just links to the 'Beginners' board), BUT...
'Copy/Pasting' the link below, excluding the brackets ("[...]")at the ends, into another browser window's directory input & hitting 'enter' opens the thread fine.
[ www.eevblog.com/forum/beginners/icl7107-volt-meter-hacked-to-an-ammeter-w-2-5v-ref-in-lo-offset-(solved)!/msg5532997/#msg5532997 ]

4. Current Sense Output Disrupting Ammeter Powered With Most SMPS But Not All(? ? ?)
https://www.eevblog.com/forum/beginners/current-sense-output-disrupting-ammeter-powered-with-most-smps-but-not-all()/

5. Safe & Simple Way To Convert Floating Wall Wart To Grounded?
https://www.eevblog.com/forum/beginners/safe-simple-way-to-convert-floating-wall-wart-to-grounded/msg5541505/#msg5541505

[AtlanticSurfer]

Thanks Trurl, seems you've already beaten a path down this route, I'll see if I can follow it. 

[Trurl]

If you are getting a negative current sense voltage(CSV), and it eventually becomes positive with loading, then the same voltage to this initial negative CSV (after it has settled with the PSU warmed up) needs to be applied to the "In Lo"(- input) of an ICL7107 based volt meter's differential input, while the CSV output from the PSU would be connected to the "In Hi"(+ input) of the meter's inputs.

If all goes as expected, the ICL7107 should process this negative "In Lo" voltage as the "zero" level(negating the negative offset), so that the CSV is "counted" from actual loaded output.

If you have not fiddled with an ICL7107 or a panel meter based on it, it may take some time and effort to get familiar with its peculiarities, but its datasheets are VERY informative(and it's much "simpler" to fiddle with than most modern chips with fancy multiplexing etc. such as those tiny V/A meter combo doodads). I learned SO much. And in my suggested threads above, I spill all the beans... 

[AtlanticSurfer]

Thanks again Trurl, I’ve read through all the threads you linked to (and more) and I’m sure if a chose to go down the ICL7101 route that your thorough documentation would make it much easier for me to get a result but the learning curve seems too long and too steep for me at the moment.

It occurred to me that I might be able to use the atmega32u4 MCU’s that I have in my drawer.  They have six analogue input pins that can read voltage down to -0.5V and I believe that a differential measurement may be as simple as subtracting one pins voltage from another’s in code.  Presumably that measurement can be used to output a voltage to the cheapo meter in place of a shunt.  After reading those threads today I think I’d power the MCU and the meter by battery to keep things simpler.   I’ll have a dabble at that.  The MCU route would leave the door open for controlling digital potentiometers for voltage adjustment and from there remote sensing.

[Trurl]

You obviously have more sophisticated means at your disposal. It'd be a treat to read about your experience. When your solution gets up and running, please do brag at will with fancy photos and all, right here in this thread for all to be blown away. Cheers~

[Gorbatiuk]

Hi.

I finally bought myself this modified PSU and I think I figured out how it's made on this board with jumpers.
It's a feedback circuit, so it won't work with just resistors. As far as I understand, it takes +12V from two points, one point to the shunt, the other after. It goes through the current sensing amplifier INA180A4 and feeds the output to the non-inverting input of the operational amplifier LM321. +5V is applied to the inverting input through a voltage divider (the value is set by a jumper). The output through a diode is fed to the same place where the voltage is set. So, the higher the current, the higher the voltage at the output of the LM321, and accordingly, the voltage and current at the output of the PSU will decrease, and vice versa.

Schematic of first PCB:
(Attachment Link)

The function of the other board on the large capacitor is still not entirely clear. It has a 4-channel operational amplifier LM324 and an optocoupler for power control, with wires from it going to the power-on pins. This board might also be responsible for reducing the fan speed. Additionally, the device turns on and off in an unusual way, as if it's going through a protection trigger (there's an audible beep).

Regards.

Hello, Evgeniy.
Could you please tell me where the OVP resistor is connected in the unit you bought and what is its rating? Thanks in advance.

[zheka64]

Hello,
I didn't look inside mine because it was all taped up and I didn't want to mess with it. In my other PSU, I soldered a resistor like in the picture.


There's also a resistor there, probably for undervoltage protection. It's normally a 20kOhm resistor, which sets the cutoff at about 11.8V. The modified PSU I bought had a 10kOhm resistor, which they say sets it to 7.5V. I soldered a 100kOhm resistor over the existing one 20kOhm in my second PSU, giving a total of approximately 17kOhm and a calculated cutoff of 10.4V.

[Trurl]

zheka64, were you able to notice any differences/advantages to placing the 4.7k OVP hack resistor at the location you indicated above? I can tell it would be a bit easier(more spacious) than the original location(very tight space) as shown in Hopper's video, but I'm wondering how this other location for the OVP resistor affects the PD43's performance behavior(if at all). Perhaps a whole different set of values for OVP would have to be assessed?

Also, please clarify how "a 100kOhm resistor over the existing one"(10k), gives "a total of 17kOhm". Thanks...

[zheka64]

Hello
Trurl, the device works normally; OVP does not trigger in the range of 12.3V to 14.6V, and I haven't checked above that. A 100kΩ resistor in parallel with the factory-installed 20kΩ resistor results in 17kΩ or 16.667Ω, to be more precise. I updated my previous post to make it clearer.

[AtlanticSurfer]

trying to make a power supply contained in a hammond 1455T2202BK enclosure.

Nice job enclosing the unit!  My enclosures are due to arrive at the end of the week and I was wondering if you might be able and willing to share the cut-out dimensions for the the endplate? 

[Trurl]

For anyone intending to try the "alternate" OVP placement location, as demonstrated by zheka64 in the altered version available in Ukraine(with added PCBs & current limiting etc.), please note that this "alternate" OVP resistor location affects the voltages on either side of resistor "05C(2)", as marked in my schematic, slightly differently - according to a simplified simulation I performed in Proteus software [i.e. the voltage at 05C(2) resistor's lower solder point - the original OVP hack point - and the upper solder point, will be different when the OVP resistor is placed on the "alternate" location] so it is likely that my uploaded chart with OVP trigger characteristics and its resistor values will not transfer over to this "alternate" OVP placement(i.e. this "alternate" OVP placement location will have its own slightly different OVP trigger characteristics and associated resistor values - particularly without any of the Ukrainian version's additional PCB addons & wiring).

[AtlanticSurfer]

For anyone intending to try the "alternate" OVP placement location, ...

I tried using the alternate OVP point but ripped the pad off the board when tidying the lead - it's very fragile.  If trying it again I'd attach to the upper solder point of the 05C(2) resistor since it has continuity with the alternate point but is more robust.

The unit seems to function okay without the pad after reverting to the original point.