Fixing my Mastech HY3005D-3

[TeraHz]

Hi guys, I've had this PSU for few years and recently as I was charging my lipo batteries (via charger, not directly) on both psu outputs at 17.8V @2A, my PSU went berserk and flipped one of the outputs to 50V and the other to 0. Needless to say, the charger on the 50V output died.

So I changed the input diodes, the power transistors, a bunch of caps (while I was at it) with good brand ones and managed to bring one of the channels (I think the secondary) back to life. It lost some of its upper range (only goes to about 26.4V), but I'm guessing because it needs recalibration from the new parts. However the other channel lost quite a bit of its upper range, doing 0V-0.02V

I've checked the voltage multiturn pots and they are still doing their 0-6.8kOhm range. Changed almost all of the transistors and regulators. Checked the big power caps, the diodes etc. I can't find what's busted.

I'm attaching the schematic of the PSU with some test voltages that I've done. I tried to trace the control signal from AC to the transistors which control the load. The green numbers are from the working channel, the red from the non working, as you'd expect.

I'm sure a better educated than me person can probably spot the issue right away, so I figured I'd ask.

Thanks

[Paul Price]

Looks easy enough for me. Need the V readings on the of the schematic:  N2, V22, V21, V13, V14  to give you a verdict

[TeraHz]

Here are more measurements.

[Paul Price]

I think I've localized the problem to a few parts.  V13 is the most suspicious, with V14 coming in as a close second.
Check the base-emitter, base-collector junctions carefully.

If they somehow seem to check out ok, I want you to momentarily connect a 1K resistor to the + output of the bridge supply to the base of V13 and look at the output voltage of the supply.  Whadja get?

[TeraHz]

Ok, just waking up . The two transistors check out. Desoldered them and tested them (I had already replaced them).

By bridge supply are you referring to the 7812?

[Paul Price]

The 18V measured high power V7-V10 bridge + point.

[TeraHz]

With 1K resistor to the base of v13, the relays all click and the output voltage goes to 40+.

[Paul Price]

Ok, set  output voltage control to 1/2 range of travel. Connect 1K resistor from v5 cathode to pin3 of N2

[TeraHz]

Nothing happens. Tried from 12V as well. Dead opamp?

[Paul Price]

If you disconnect V16 and the circuit behaves the same, the op-amp is kaput.

[TeraHz]

V16? That's the CC mode led connected to the output of N1?

[TeraHz]

Removing V16 makes no difference, other than when I turn the CC pot all the way down it doesn't enter CC mode.

Removing V15 (the CV led) sends the supply to 42V+

[TeraHz]

Well, it is not N1 or N2. Just swapped them between the channels and nothing changed.

[TeraHz]

I think I figured it out! The schematic isn't 100% match with the board, but it was very close. There is a big resistor on the board, which I don't see on the schematic, which appears to be between RP4 and the 0-32V line (what I wrongly called GND in the first post). This resistor was about 15kOhm on the non-working board and 470Ohms on the working board... no idea what happened there but it would explain why my output had such a tiny range. The multi turn pot is 10K so adding 15K to that line would mess things up. Changing it to 470 fixed the output. Now to do some more testing in the full range with a DMM see how accurate everything is and to put this piece back together.

Thanks Paul for the time and help!

[Paul Price]

I cannot understand the your explanation on how the resistor values changing from 4700 to 15k would  limit the output close to 0, and you talk about something way out of these values "470", but congratulations, it works! Perhaps the resistor was  actually closer to open-circuit in operation.

Success needs no explanation.

Better to test the output with a real load (and watch the action on a scope if you have one) before you put the screws back, instead a just a quick voltage test using a multimeter.

And before the screws go in, what about the other channel?

You're welcome, glad to be of help!

[TeraHz]

I changed it from 15K to 470. Not from 15K to 4.7K. I think the issue was that the circuit works over 500-10Kohm range on the  multiturn pot. A 15K resistor in series with that would make it ~15k-22K range instead, so turning the pot makes no difference it is always at its lowest voltage. Does that make sense?

Both channels work now. I'll be testing with load and will watch the scope. Unfortunately have to come up with something to do a proper load. I don't have an electrical, or even a CC dummy load.

[Paul Price]

Test Load Ideas for your power supply.

(1) One or more 3 to 10 watt 1 to 10-ohm resistors placed  in a can filled with ice and water...good for several minutes of entertainment. I always enjoy watching the tiny bubbles.
(2) A resistor from collector to base of any power transistor, even a small one on a large heatsink (a bench vise will work clamping any flat surface of the power transistor to a large piece of flat metal) will make a variable load you can use until your workbench catches fire or your little brother keels over from the smell and the smoke.
(3) Even T0-220 NPN or even PNP power transistors or power n-chan MOSFETS mounted to a 2-in sq. heatsink will survive many minutes when biased to handle 3-amps or so(100 to 1K from collector to base), if submerged in a robot's cocktail, made with plenty of ice and served in an old deep dinner plate. add enough water and chopped ice and simmer until done. Use a small PC cooling fan if you run out of ice.
(3)You can usually find one or two 3-15amp  High Voltage NPN power transistors aleady mounted  on a large heatsink in discarded apparatus, such as a discarded PC power supply which you can ravage and just rip the parts you want out quick with handy tools, for example, a chainsaw. If your pets are bothered by the noise, use use a small hammer, a large screwdriver and, if you like to get the job done fast, a hand held saber saw. The fan on it will cool the load if you power it with 12V from the power supply channel you are not testing.
(4) Car 12V lights, the higher wattage ones used in headlights or tail lights are both durable and blindingly good loads.
   

[Teneyes]

Hello, I picked up a HY3003-D3 on Ebay, and it was Blown apart, I ended up just paying shipping.
 I found debugging was better when voltage measurements were made using the Positive output (X2) as the reference point, Note that all the control logic is riding on top of the Pos, output.  In my PS the big Caps 4700uf were gone (cut out), but you may wish to also check C20 ,470uF , if you see(meas.) AC Ripple.  I my case I added a 3rd 2N3055 and convert to my HY3003D-3 into 5 amps like yours. See pics . I used a  bank of 100 watt ac lights (old style not LED!) for load
Is this a Better Schematic

[Paul Price]

It appears that the emitter of V26A wiring if definitely in error in your schematic.

[Teneyes]

It appears that the emitter of V26A wiring if definitely in error in your schematic.

  Yes Missing the connection, emitter of V26A to the output.  It was just one on the many I found.,  It was a fun mod for me, and learning experience

[Russ314]

I'm having trouble with this same unit. Can anyone tell me how to increase the current output? One channel outputs 5.7A, but the other only outputs 4.6A. Probing around, the "bad side" has a higher Vce on all the 2N3055 pass transistors, so I'm assuming I need to increase the base current somehow. The zener V11 also has a higher voltage drop on the bad side. I tried reducing R4 but that did nothing. Would a lower voltage zener help? I guess I can't really follow the circuit well enough to figure out what's happening. Maybe if the op-amp output a higher voltage, it would allow more into the base of the 2N3055's?

Anyway, as advanced payment, I'm going to teach the forum how to add fine adjustment control to this style of power supply. Simply remove the board-mounted trimpots and replace them with panel-mounted trimpots. I'm using a 10k pot for the voltage (which makes the main voltage knob go from zero to 5v, instead of zero to 30v), and a 500k pot for the current (which makes the main current knob go from zero to 100mA, instead of zero to 5A). With this simple modification, you can easily adjust the current to a tenth of an mA, which is essential for doing LED testing where you want to set it to exactly 3.1mA, or 8.7mA.

[Russ314]

I just realized the schematic in OP's post is quite a bit different than mine. My unit appears to be a knockoff (what the heck is HE Test?), but it's 15 years old, and the layout is identical to the Mastech original. As such, I've attached the schematic from Mastech's website. Notice it has the three 2N3055's in parallel, so they all have the same Vce. I figure I need to either increase the base current, or maybe replace the transistors... which I'm trying to avoid. The unit doesn't even get warm when I power a resistive load at 10A for several minutes, so I feel like I can definitely squeeze more current from it somehow. And I've never used the little standalone 3A output, so I'm hoping that means the main transformer isn't saturated. It'd be nice if I could squeeze that 3A from the main channels somehow.

[Teneyes]

Hi Russ... . I has been awhile since I modified my PS , and here are some comments.
In your testing at 10 amps .did use an independent ammeter?  .
It does take awhile for the heat sinks to warm up at 10 amps. 
   Reviewing the schematic it appears the adjustment of W5A, W8 pots set the limits of the current adjustments, along with W6 (front panel pot).  You may check those.  Also W3
For fine adjustment of voltage , I replaced front pot with a 10 turn Pot ,Voltage only.
I have a un balanced current situation in my PS (parallel mode) that I am looking into Also.
Good luck to you.

[floobydust]

I would do a current-limit calibration, three trimpots. Unfortunately, they all interact with each other and I've forgotten the sequence, so there is some back and forth and experimenting needed.

Trimpots W8, W5A interact the most and also changing front panel pot W6 (end-end value) would upset the CC calibration. I think how it works:
W3A sets the CC op-amp's offset when not in CC mode. In CV mode at minimum output voltage setpoint, I guess set W3A just above the point the CC LED lights or maybe around 0V at pin 6 is the goal. I thought it can also upset priority whereby CC does not overrule CV which is bad if you short the output.
W8 sets the PSU's maximum output current. Dangerous to go way over the design's max. for a long period of time = smoke.
W5A sets the PSU's minimum output current. I think a few mA is the best it will perform.
W6A sets the PSU's maximum output voltage.

[Russ314]

For fine adjustment of voltage , I replaced front pot with a 10 turn Pot

Yeah I did that first, but wasn't happy with the results. That reminds me though - for anyone having strange intermittent problems, try adding conductive nickel ink to the rivets on the potentiometers. My NEW replacement pots ended up giving me problems on cold days until I coated them with nickel (using an MG pen). I can't believe that style of pot uses rivets to mash the pins directly against the carbon traces... what a joke. If I had to do it all again, I'd definitely use rivet-free pots.

I have a un balanced current situation in my PS (parallel mode) that I am looking into Also.
Good luck to you.

Thanks for reminding me! That's another symptom; very uneven current in parallel mode. If I set it to 5A in parallel, I'll get 1.5A from the weak channel and 2.5A from the strong one.

[Russ314]

W8 sets the PSU's maximum output current.

W8 doesn't actually exist. I've literally never seen one of these units that had W8 inside. And maximum current is set by setting all the trimpots to zero, i.e. short circuit. In other words, doing anything whatsoever with the trimpots won't increase the output, it will only decrease it.

What about decreasing the 3.9k resistor directly above W5A? If a higher voltage goes to pin 3 of the opamp, it would be like setting the trimpot to below zero, right? The only thing that bothers me about changing resistors is that I've checked literally every resistor on both boards, and never found a discrepancy from one side to the other.

P.s., I just checked the currents with an independent meter and the display values are correct.



**UPDATE**

Wow, reducing R15A actually worked. I probed in some parallel resistance to that 3.9k resistor and the output jumped from 4.7A to 5.9A. The weird thing is that after removing the resistance, the output stayed over 5A. Clearly, heating some part of that circuit is increasing the output. I know this is probably not the best way to do this, but it's good enough for now.

I'm still curious though - what is the actual limiting factor in this design? Why was 5A chosen as the output limit? Three 2N3055 transistors in parallel can obviously pass a heck of a lot more than 5A.

[floobydust]

When it wimps out on current, is the CC LED lit or not?
If not, you could lower R5A from 100R to 50R 1W for more base-drive to the pass transistors, if they have low hFE and that is limiting things.
If the CC LED does light, and there's no trimpot W8 1k, just a jumper? A 10k main pot W6 for current adjust and W5A 5K trimpot?

I measure potentiometers end-end because they are all over the place, tolerance ±20%.
The Mastech schematic says 6.8k and 4.7k, part values that do not exist really.
I would put in the trimpot, something is needed to cover tolerances in parts.

I thought highest voltage at IC pin 3 will give highest output current, so W8 even a 500 ohm resistor or lowering R15A 3k9 would fix that, so you can dial in 0.22V at least. Emitter resistors are 0.22R X3 so 0.22V/A output each. 3A would be at least 0.22V for pin 3. I don't know the mods you've done.

I believe the (current) limitation is the SOA for the pass transistors and their heatsinking. 2N3055 is pretty cheap as an Asian part and they can have low gain. Steady-state they look way underutilized, but shorting the PSU's output overstresses them.
Some units I worked on, the driver transistor 2SD313 is the bottle neck they tend to overheat as well.
The voltage drop across R5A will tell you the base drive current for the set of three.

edit: math bad

[Russ314]

When it wimps out on current, is the CC LED lit or not?

Yes, the CC is lit. Although in parallel mode, both LEDs are lit at the same time, at half brightness. I've never known if that's normal or not. I've attached a modified schematic that shows my layout. I'm not sure if the 3.9k is actually 2.15k btw; that's just what it measures as, in-circuit.

I believe the (current) limitation is the SOA for the pass transistors and their heatsinking. 2N3055 is pretty cheap as an Asian part and they can have low gain.

If I remember correctly, the transistors say"TransSun" or something. I'm not sure I want to spend the money to buy real ones, but I've always wondered if I could swap them for something else. I've got a pile of unused 2SD2560's here that seem like good candidates. Could they be directly swapped in? They have a much higher gain, that's for sure. I've added a pic comparing the two.

Steady-state they look way underutilized, but shorting the PSU's output overstresses them.

I'm definitely not one of those people who shorts the outputs to set the current... I always drop the voltage externally through a massive low-value ceramic resistor. I do, however, do a lot of high current PWM'ing... I think my problems all started when I was PWM'ing an 8A motor at 20kHz.

Some units I worked on, the driver transistor 2SD313 is the bottle neck they tend to overheat as well.

I've always wondered if I should put a larger heatsink on that part. I've upgraded or added heatsinks to every To-220 package except that D313. It has a rather pathetic stamped heatsink from the factory. Then again, I've never felt it get hot. For how I use this PSU, I never run it for more than a minute or two at a time. I use it for development, not anything steady-state like battery charging. In fact, in 15 years, I doubt if I've ever had it run for more than 5 minutes straight.

[Russ314]

So my Digikey order arrived and I spent the ENTIRE day replacing parts, upgrading components, and... failing to resolve the problem. I've attached pics to demonstrate a main symptom of the problem. When I drive a resistive load in parallel mode, the current and voltage lag on the slave channel. What gives?!

[Russ314]

Just made in interesting observation - the values all change depending on how you plug the cables into the front. I'm going to remove the front section next and investigate the diodes/caps on the output jacks.

[KG7AMV]

Thanks.

[Russ314]

Here are some things I've learned, for the benefit of future lurkers:

1. The 2N3055 pass transistors are basically ideal, but if you want to upgrade them, make sure the replacements have a TO-204AA, or TO-204AE, or TO-3 package (using Digikey's parametric search). Also make sure the voltage/current rating is 60V/15A or higher, and the manufacturer is ON Semi. If you input those parameters, you'll get 19 search results, ranging in price from $5.85 (for the 2N3055) to over $15!

2. I removed resistor R15 (which ended up being 3k) and replaced it with a 2.2k, which increased the peak current output from 4.7A to 5.5A. This proved to be a quick and viable way of increasing the current output.

3. I discovered an empty trimpot footprint on the top left corner of the board, labeled W4. This proved to be the missing W8 trimpot, according to the schematic. At first I installed a 50k trimpot in this location (which was the ideal value), but then I decided to remove it, because I was worried about unnecessarily coupling ground noise to the op amp.

[Russ314]

EDIT:  Sorry, W3 (the offset trimmer) is actually to set the voltage zero. In other words, if you hook up a sensitive volt meter to each channel, and turn every knob on the front panel to zero, you can adjust the trimmer until the output is 0.000V. This will take some fiddling, because it likes to jump straight from -0.4v to 0.4v. Getting the output to read exactly 0.000v is almost impossible. And FYI, when the output is negative, the front panel of the Mastech will read 0.00v. So don't be surprised if you go to verify that "zero volts" reading it and find that it's actually negative.

P.s., I still haven't solved my problem, but I have discovered that when PWM'ing a 5v, 1A load at 32kHz, the output ripple is 0.385V, which equates to 8%.. I was hoping for better than 5%. The ripple was identical on both channels. Is this an output capacitor sizing issue?

[Russ314]

In my final configuration, I replaced R15 on the slave side with a 2k, 1W resistor. This made the peak output 5.8A on each channel. However, in parallel mode, I'm getting 6.5A on one channel and 4A on the other... I haven't turned it higher than that, because the 6.5A scares me. All I want is 5A from each channel in parallel mode!!

P.s., I changed the output capacitor from 470uF to 5,400uF as a random test, based on what I had lying around, and it reduced the ripple from 8% to 3%.

P.p.s., I added 0.1uF/50v/X7R caps all over the board, across the rails at the op amps and linear regs, and it had zero noticeable/measurable effect.

[mijudesign]

I have the same HY3005D-3 power supply, which i owned since new


Issue
Voltage would fluctuate and the relays would click frequently

Steps Taken
replaced the main capacitors (63V 4700uF) and the regulators (2N3005)

Outcome
Voltage DISPLAY is very stable

Issue
Extremely sensitive control dial, wild fluctuations when touched

Steps Taken
Turn the control dials back and forth 30 times

Outcome
No longer sensitive, smooth control

My question to the forum, where it seems many people have adjusted and repaired this device previously, what are the recommended replacement parts for ongoing maintenance (minus the parts above) ?

I presume with my very limited skill that the low volume hum from the power supply can be improved if I keep changing capacitors but I was very keen to have confirmation. All inputs appreciated.

[Teneyes]

Hi,
   I replaced one of Voltage controls with a 10 turn pots, for those times when I like fine adjustments.

    Replaced bridge diodes too. 10amp.

    I am thinking of  making a modification to the relay step up point so it is not near 14Vdc,  to be not near the 12V charging / float levels of 14.4 - 13.2V for,rad acid batteries

[mijudesign]

Im interested in these upgrades too. Do you recall the part numbers ?