Heathkit IM-2410 frequency counter power supply problems

[SingedFingers]

Managed to grab a Heathkit IM-2410 225MHz counter from eBay. Inevitably for the age it has a couple of problems with it but it mostly works. I am however stumped on one of the problems.

The power supply isn't working very well. The schematic is as below. The 12v rail is regulated by a TO-92 78L12 and the 5v rail is regulated by a uA7805 (1.5A variety)



Quick voltage check showed low voltages on both. I expect a volt or two less as this is a 50Hz country and it's a 60Hz design ... The 11v point is reading 9.5v and the 20v is reading 17v. These are within the dropout specifications of the regulators. They were a LOT lower than this but I replaced all the electrolytic capacitors with Vishay/BC 021 long lifes as they were rancid old TI and Nichicon branded things from the 1980s.

The capacitor replacement killed a number of ills. However periodically it refuses to count for a few seconds. I decided to pop the scope on the 12v and 5v rails after the regulators and have a look on the basis that this might be dropout causing the input bias to get screwed up which makes sense for when I have traced it doing it. Both of them have a rather large dip in sync with the smoothing caps charging. See the trace below.

The bottom trace is 1v/div taken from the output of the 5v regulator. This is peak 5v dropping down to 4v.

The top trace is 2v/div taken from the input of the 5v regulator. Peaks at 11v, drops to 9v.



I've never seen R30 used in a regulator topology before and it's a rather large 5W carbon composition 20R resistor in this built version, not a 10 specified in the schematic and it gets quite warm.

Annotated board:



Note: BOTH the 12v and 5v regulator outputs have the bottom waveform which is confusing me.

Note 2: transformer is humming slightly.


Any ideas what to look at next or any diagnostic approaches to try?

Any help appreciated!

[SharedKnowledge]

I assume that's a 100hz waveform (full wave power line ripple)? That's important to know as it could also be the counter circuitry drawing a lot of current at some other repetitive frequency.

It's likely the total load on the power supply is excessive for some reason. If the counter is working it's hard to say what could be wrong but you might look for components that are running warm, leaky capacitors, etc. You might also measure the actual current draw after the 7805. If the counter sometimes doesn't start up in the first few seconds that might also be a sign of leaky capacitors or excessive initial current draw.

R30 appears to be a bit of a "hack" and is likely there to help the 5 volt regulator not overheat and/or current limit. It's also a carbon comp power resistor which are famous for radically changing value with age especially if they run hot. So I would pull one end of that resistor and measure its real value. If it's more than 20 ohms you may have found your problem. If the resistor measures okay you might try adding a 4700uf or so filter cap in parallel with C36 to remove more of the ripple. Running on 50hz you may need bigger caps in the power supply. You might also try a slightly lower value for R30 and use a 5 watt wire wound type.

I should add because of R30 it's important to keep some minimum load on the 5 volt supply or the voltage will go above 5 volts and potentially damage whatever is still powered from it. Likewise reducing R30 to a lower value should be done cautiously. The load on the 5 volt supply likely varies with what numbers are displayed and possibly with different settings.

[SingedFingers]

Thanks for the info. Some replies:

Yes that's 100Hz.

I've measured after the 7805. Fortunately it's on a nice little connector and attached to the bottom of the case.The total draw is 1.1A when counting at 148MHz. It doesn't get that hot which is good at least.

R30 has been removed and measured as well - it's bang on 20 ohms. I don't like carbon comps so it'll be replaced soon

I have clipped another cap across the original one that was in it (another 4700uF I have lying around) and that made no difference.

Very annoyingly it's been sitting in front of me for the last hour and hasn't dropped count even once!

Edit: measured temperature of the 7805 and it's sitting at 38oC which is fine. Resistor sits at 51oC.

[xrunner]

I doubt this is the issue but I had an old Keithley DMM that had issues like that. It turned out that even though the capacitor was soldered in properly there was a broken trace that prevented it from filtering. I'm sure you have already done this but I'd inspect the trace side of the board carefully. 

[SingedFingers]

Yeah I've fallen down the shoddy soldering hole for a couple of hours before so checked everything first

Detailed account of shoddy soldering in commercial gear that I encountered here: http://www.vintage-radio.net/forum/showthread.php?t=131243

Anyway it appears to suddenly have sprung into perfect usability. It's been sitting here for over an hour like this without a single damn glitch:

[Seekonk]

I had one of those I bought used long ago. It was a crappy design at best and ran really hot.  I bet you still see a little drop in the 5V.   I added extra capacitance to mine with little improvement.  If I owned it today, I'd put in one of those little buck converter boards.

[SharedKnowledge]

It's also possible, but not likely, the transformer has a bad winding on one side of the center tap. Roughly doubling the filter capacitance should cut the ripple and, if not, that's suspicious. It's also possible the 5 volt regulator has an issue. The TYPICAL dropout is specified at 2 volts at 25C and 1 amp. It's running over 1 amp and over 25C both of which is going to degrade the dropout performance. The TI datasheet doesn't give a max (worst case) value for the dropout spec. You're also using an analog scope and it's hard to get exact measurements so it might be dipping lower than 9 volts. And if you're using AC coupling on your scope that makes the measurement even more suspect.

The dropout waveform on the output is not what I'd expect in the scope shot you provided but that might partly be due to R30. I would expect the regulator to recover quicker if it is dropping out.

And if you're saying the 12 volt supply has similar drop outs, that seems to indicate the real issue is the 50hz power. It might be the transformer is saturating.

[SingedFingers]

I've got a few ua7805's floating around. I might swap it out and see what happens. I imagine as these are much newer devices (date code is 2003) the dropout voltage might be a little better. I've seen terrible drop out on older regulators. The transformer does worry me a bit. It may be replaced as I have a suitable one already with a 18v 50va rating that will (just) fit in the box. That's about as overrated as it needs to be

Seekonk: it's TTL - it always runs hot and helps keeps the house warm! You also can't stick a switching converter in it due to the switching noise triggering the front end.

I've actually got an even older IB-1100 stuffed full of nixies and 74 logic (not even LS!) arriving in a couple of days as well. Should be interesting, and hotter.

[Seekonk]

You also can't stick a switching converter in it due to the switching noise triggering the front end.

Naw, the world is full of digital stuff and it works.  You've been listening to regurgitators.  That front end has to be hit with a lot of signal.  Had a nixie one too.

[xrunner]

Anyway it appears to suddenly have sprung into perfect usability. It's been sitting here for over an hour like this without a single damn glitch:

Yea I know how it goes - it's simply waiting to fail when you really need it. 

[SingedFingers]

You're probably right. Equipment does that from experience 

Seekonk: I've designed and built two frequency counters from scratch before. It's not quite as easy as that. Trying to get stable automatic hysteresis at 100MHz+ is a bitch when there's spikes galore on the rail. Even if you filter out everything with LC Pi filters, decouple, isolate, stick behind shielding, docouple some more, it's a big problem at mV sensitivity levels. My current "homebrew" one is good to 120MHz with 5mV sensitivity at that frequency. Unfortunately it has software problems I can't be arsed with (day job is software) which is why I bought this one.

[james_s]

The resistor is a bypass to allow the regulator to carry more current. It's a bit of a kludge but I've seen it before, the regulators on the HV board of the Amplifone vector monitors used in a few Atari arcade games have resistors like that.

It's possible that the transformer is not happy on 50Hz. You can pretty much always use a 50Hz transformer on 60Hz but the opposite is not always true. The core of a 60Hz transformer may saturate on 50Hz and the VA rating of the transformer will be lower.

[SingedFingers]

Thanks for the reply.

I've decided to leave it as is for now. If there are any problems in the future I will be doing the following mods:

1. Replace transformer.
2. Wire an LM317K in as the 5v regulator
3. Have some hot nibbling action and stick an IEC socket on it. I hate integrated cables.

[hamster_nz]

Assuming the regulators are good (they cant both be identically bad), the only reason I can see for the output voltage dropping  rather  than rising as the charge up  is that GND pin on the regulator is somehow solidly held  at 0V.

Have a look between the center tap on the transformer and the GND pin on the regulator -  there should be nothing there....

[SingedFingers]

interesting idea. Will measure that later.

[beeeerock]

This thread is pretty old, but I'm wondering how this all turned out?  I was recently gifted one of these counters and also noticed the ridiculous amount of heat it produces.  In my case, it's virtually all from the 8 LED driver chips... and the 20 ohm carbon resistor at the back (the one that's supposed to be a 10 ohm wire wound according to the ancient manual!).  I don't know if this is as good as it gets and I should just live with the heat, or if there's something wrong that I can sort out.  I have noticed that the 5 V supply is only 4.75 V, which is apparently right at the bottom of the range on the driver chip spec sheet.  I'd love to know why the resistor was changed (apparently) after the manual and schematic was printed.

The drift of the counter was pretty extreme, likely due to the excessive heat.  I ran it with the lid off to help with that (it did improve to some extent), but it still wasn't solid enough for my needs.  So, in case you're interested, I cobbled together a tiny TCXO synth board with an Arduino to provide the colorburst frequency to the 8 pin divider chip.  It works remarkably well and the result is a stable counter that can be calibrated very tightly.  I've ordered a ProgRock2 from QRP Labs to replace the mess I have now and will be able to add a 1pps GPS signal at that point.  Crazy to think this old clunker can be accurate!