DSE D-3800 power supply fault

[VK7HH]

Hi All,
I've got an interesting fault with my DSE D-3800 P/S unit.

All works fine normally, no problem with regulation/current etc. However, sometimes the unit will be switched on and there is no output. Both the Warning (Overload) and ON LEDs are lit. There is no short circuit on the output.

Putting a heatgun on the side of the P/S for about 10-15 seconds causes the output to reappear again. I've opened the unit up, resoldered some suspect looking joints, tested with cold spray/heatgun to see if I can trace the location of the fault to no avail. It's most frustrating when your part way through transmitting and the output fails.

Anyone got an suggestions or come across this issue before?

Hayden

[VK3ALB]

Hi Hayden,

Intermittent faults can be tough to track. You could take some measurements in the good state and compare them when the thing stops working. If you don't have a copy of the circuit you can download it from mods.dk.

[VK2ZIP]

Hi Haydon / Lou,
I have the same problem with my D-3800. It has happened intermittently for several years now, but I have never tried to track it down until now.
It was solidly "off the air" last night, so I opened it up. Then I found this post on AUSTECH FORUMS
http://www.austech.info/showthread.php/9...elp-needed
which gave a link to the circuit diagram, information in the thread saying that the main control IC is an LM723, and some other helpful suggestions.
I heated the main PCB with a short burst from my heat gun, and it came to life straight away this morning.
I will measure pin voltages on the LM723 in normal operation and post them later.
In the meantime, can you tell me if you have made any progress on this problem?
Best wishes,
Lew.

[VK2ZIP]

Hi Haydon / Lou,

The attached document shows LM723 (IC1) pin voltages for normal operation at outputs of 10 V and 14 V, i.e. when the supply is working properly.

It's a pretty standard configuration driving a composite Darlington configuration consisting of a TIP31C feeding 5 x 2N3055s in parallel. You can find examples of this configuration on the internet, but there is one important difference. The Frequency Compensation pin is being used for Remote Shutdown, via a diode from pin 13 of IC2D to the frequency compensation pin of the LM723 (yes, pin 13 on both ICs.) I have a strong suspicion that the fault is actually related to IC2.

Best wishes,

Lew.


  D-3800_LM723_Normal.pdf (Size: 41.64 KB / Downloads: 887)

[VK2ZIP]

Hi Hayden / Lou,

I have made some progress, but not all of it is good!

My D-3800 went back into fault mode shortly after I posted the (IC2) LM723 "normal" pin voltages, and no amount of heat from my heat gun would persuade it to recover, so I have not been able to measure normal voltages for IC2, which is an LM339 quad comparator.

What I have ascertained, however, is that as expected the problem is in the LM339 area. When I disconnected one end of D12, which is connected from the open collector output of IC2D to the Remote Shutdown pin of the LM723, full output is restored. The D-3800 works perfectly, but of course without the protection afforded by the LM339 circuitry. That is not to say that the unit is totally unprotected - the over-temperature fan still works, and presumably the LM723 current limiting still works also, but I have not analysed that yet.

Hayden,

If your problem is in the LM339 area, this "fix" will work for you too, but I would not want to be responsible for suggesting it as a permanent solution. Lifting one end of D12 will be useful, possible indispensable, for diagnosing the problem however, as it restores power to the LL339 area, so that you can check voltages. I have been doing that, and so far I can only say that I am confused!

Best wishes,

Lew.

[VK7HH]

Hi Lew,
My power supply hasn't been playing up as much since I originally posted, just noticed your replies here. Today it failed again, after running for almost a month non stop.

I also posted a similar query on the Ham Radio Australia page on Facebook, here is some of their advice. I've yet to pull my P/S apart again. Electrolytic are the first place to start.
http://www.qsl.net/pa3fqh/ep920/EP-925_Schematic.pdf
http://pa0fri.home.xs4all.nl/Diversen/EP...925eng.htm
http://www.qsl.net/pa3fqh/ep920/ep920.html

[VK2ZIP]

Hi Hayden,

Thanks for your reply, and for the 3 very interesting links, which contain a lot of useful information and some worthwhile modifications.
It is particularly good to have the redrafted circuit diagrams - I could not reliably read all the component values on the circuit from mods.dk.

Since my last post, I have managed to get hold of a working D-3800, and I have measured all the pin voltages on IC1 and IC2, for output voltages of 14 V, 10 V and 5 V. You will find those in the attached PDF.

When you get around to opening up your D-3800 again, note that it has a number of components not shown in the redrafted circuit diagrams, specifically R41-C26 (in yellow ellipse), an RC filter across the inputs of IC2D, and D16-D17-R42-R43 (in red ellipse - D16 is in the white sleeving), which modify the signal to the + input of IC2D. These components are only shown in the circuit from mods.dk, and are all soldered point-to-point on the solder side of the PCB, as shown in the attached image. (Note that C25, which is soldered between pins 7 and 10 of IC1 on the solder side, has been removed in this image.)

I am now fairly sure that the fault in my D-3800 is IC2D (LM339). It's output (pin 13) is low even though its IN+ (pin12) is greater than IN- (pin 10). This is shutting down the LM723. As indicated in an earlier post, lifting one end of D12 restores full output voltage.

Capacitors are a good place to start of course. If your fault is in the IC2D area, the only relevant capacitor is C26. Connecting across C26 is a convenient place for monitoring IN+ and IN- of IC2D. I tried removing this capacitor, but it made no difference; not surprising in view of the fact that I always find the differential voltage across IN+ and IN- to be healthy (i.e. positive).

By the way, my fault mode is not always full shutdown. When set for 14 V, the supply output sometimes wanders around in the vicinity of 7 V. Full output is restored when I lift one end of D12, and then I find a wandering voltage of about 7 V on the output of IC2D (again with IN+ greater than IN-). That also seems like a fairly solid indication that IC2 is faulty in my unit.

I will replace IC2 (LM339) next and let you know what happens.

Best wishes,

Lew.
   

  D-3800_IC1_IC2_Pin_Voltages.pdf (Size: 82.35 KB / Downloads: 783)

[VK2ZIP]

Hi Hayden,

QUICK UPDATE: My D-3800 is fixed, and I was wrong; IC2 (LM339) was not faulty.

I started removing IC2 by de-soldering pin 13, and I found that the solder pulled way from the pin completely and it was no loner connected. (As an aside, this PCB is really easy to work with. The holes are slightly oversize, not plated through, and slightly chamfered on the solder side. This means that you can easily disconnect any component by de-soldering with a good solder sucker.) Before going any further, I decided to check voltages on components normally connected to pin 13 of IC2, and I found that they were still too low - 7 V instead of 15 V. I then de-soldered pin 9 of IC2, and that made no difference. Finally, I de-soldered one end of C10, which connects between IC2 pin 13 and ground. Bingo, all the related voltages increased to 15 V! I have now removed C10, and reconnected everything else, and the supply is working normally.

C10 is an 0.1 uF ceramic, 25 V. That's the last thing I would have expected to be faulty! I currently have it connected to a 15 V power supply in series with 100 k. The voltage across the capacitor started at 5-6 V, and it is now 0.8 V. This explains all the symptoms that I was observing.

I have not replaced C10 yet, and I am contemplating replacement of the 5 small electrolytic capacitors on the PCB before putting it all together. I have just noticed that C26 is non-polarised.

This seems to be such a rare fault, that you might not dare hope that you have the same fault, but please let me know if you do.

Meanwhile, the second D-3800 has failed, and it seems to be a different fault, associated with IC1 (LM723). Stay tuned!

Best wishes,

Lew.

[VK7HH]

Hi Lew,
Quick update from here. Opened up my D3800 after another failure. Poked around with the heat gun which led me up the garden path a few times. Replaced caps around the rectifier to no avail, measured volts there in the fault mode, was fine. Then replaced a dodgy looking 2N3005, still no go. Poked around and could get the supply to switch between fault and normal by blowing hot air from the heat gun and then blowing cool air from my mouth around the LED positions. Reflowed solder around these components, and couldn't fault the supply again. Put it all back together and it died again.

I'll try again tomorrow. I haven't taken the front panel off yet to remove C10 and other components you mentioned. It's a rather tedious job, and even more so to put it back on again.

Hayden

[VK2ZIP]

Hi Hayden,

Thanks for the update. Yes, pulling it apart is all a bit of a pain. I did not actually do that until Thursday.

However if you just want to test C10, try de-soldering one or both ends, or if you want to check the whole IC2D area, just de-solder one or both ends of D12. That worked for me. If it turns out to be C10, you could probably just leave it there and solder a new 0.1 uF capacitor between Pins 12 and 13 of IC2 (LM339).

C10 is the closest connection to Pins 12 and 13. D12 is not hard to find by tracing voltages and tracks from Pin13 of IC1 at the top end and Pin 13 of IC2 at the bottom end.

Good luck.

Lew.

PS I replaced C10 and the 4 component-side electrolytics yesterday, and put it all back together. It is working fine. I'm hoping to tackle the second D-3800 this morning.

[VK7HH]

Hi again Lew,
I took the front panel off yesterday. I had a play with the heat gun around the board and heating/cooling around components C17, VR2 and R17 caused the unit to alternate between overload and normal operation fairly consistently.

First I replaced C17 which had no effect. Next I replaced VR2 and tested. I was not able to get it to go into the overload state. I replaced R17 for good measure and put it all back together. VR2 is the current limit setting. I suspect the pot was dirty and going intermittent, perhaps causing the current overload protection to constantly trigger.

It's been running overnight and has been fine. I'm now contemplating adding a crowbar circuit and running the fan a different way as it's way too loud.

Hayden

[VK2ZIP]

Hi Hayden,

Thanks for the update, and well done on diagnosing a difficult intermittent fault. I think I got lucky when my intermittent fault became more permanent. I can see that VR2 going high in value could indeed trigger the overload signal, and it is a very plausible intermittent fault. How did you adjust VR2 after replacing it?

Fitting a crowbar seems like a good insurance policy. It would be interesting to see whether one could get away with something less than a 5 A fuse on the 240 V side, in order to minimise the crowbar current.

I also got lucky with the second D-3800. The LM723 had failed. I think I must have shorted a couple of pins when measuring the pin voltages. I have used a socket for the replacement LM723, and I'm replacing the component-side electrolytics before putting it back together. This time I will remember to recalibrate the Voltmeter while it is still accessible!

Best wishes,

Lew.

PS On the second unit, it was very hard to separate the PCB from the 5 mm terminal posts. I had to de-solder around the posts before it would let go. I will filing out the holes a little before reassembly.

[VK7HH]

My D-3800 died again last night... though, mostly my fault...
It suffered a short circuit on the output of the P/S.

Voltage is no longer regulated, sitting at 21V so I suspect one of the regs. I had some spares but they've been misplaced since the last fix.

Casualties:
1 x Yaesu FT-8900R (no crowbar circuit protection on this P/S). Little diode gone inside the radio so hopefully will be an easy fix... I hope... D1042 a P6KA18 transient voltage suppressor. Someone mention it's just a fancy name for a 15-16V zener.

1 x D3800 P/S

Near Misses:
FT991 - funny enough it powered down, I got the output off the radio quickly... massive panic set in... reconnected it to a battery and it powered up like nothing had happened. Phew!

[VK4ADC]

Strange, the D3800 power supply is short -circuit protected. I use one on my workbench all of the time and wouldn't like to guess how many times my flyleads have accidentally "met". No damage happens, just un-short the leads and the power supply is back to normal. Unless you disabled the over-current protection when you last 'fixed' it ???? 
 
I have my overload current pot behind the small black round panel-filler just below the voltage control knob and it needs to be adjusted cautiously.  I use long nose pliers to adjust it so that it protects at short-circuit but still allows you to power a 100W HF transceiver at close to full power - a fine line. Once set, forget. No extending the knob to accidentally change the setting either, mine ends level with the plastic front panel to avoid this.

The presence of 21V on the output sounds like a short-circuit pass transistor. If it is, you should replace all of them with the same batch code on the case so that they are all likely to have essentially the same hFE and thus share the DC load currents equally , and that is despite having equalisation resistors in the emitter leads. I found my schematic for the D-3800 and it reveals 5 x 2N3055 transistors, bases and collectors in parallel, driven by a 723 regulator through a TIP31C. It could be all of the 3055s are shorted or maybe just one.

There should always be a properly rated in-line fuse in the DC power leads to any 12V radios so that it blows if the over-voltage protection diode (virtual crowbar) needs to short-circuit. No fuse, no protection.

Advise by email or PM if you need the schematic : 1083KB PDF.

[VK5TM]

Someone mention it's just a fancy name for a 15-16V zener.

No.

These are protection devices not regulating (as in voltage regulator) devices.

For a start, once the rated voltage is exceeded, they become a virtual short circuit until the voltage drops down again, they won't regulate voltage like a Zener diode.

They are designed to handle spikes on supply rails, control lines and signal lines, not extended overvoltage events, that's why it's gone pop in your rig.

A lot, not all, can be put in circuit any way round, try that with a Zener.

[VK7HH]

Thanks guys.

I've found a suitable TVS to replace the one in my FT8900.

Doug - yes I thought the same thing. I have shorted the leads before and it's gone into overcurrent protection. The only fix I did last time was replace the pot controlling the overcurrent... it was dirty and kept tripping the supply.

Something else was up I suspect in the supply when it went.

Anyway I'm going to get more 3005's and see how I go.