Follow Up Note: At @41:36 I check the 5.2V rail one last time and see that it is still low at ~4.7V. I speculate it could be due to the lower voltages supplied by the TM5000. This turned out to be false! It was actually a faulty LM723! I replaced it and now the voltages are in spec. I may do a short follow-up vid going over some of additional repairs (including this one) required to make this SG503 fully operational. Thanks for watching!
Nice repair…good to get down into the “guts” of a “sea of TTL gates” from time to time. I was “following along on my own download of the circuit”…I did not know or pre empt the fault before you got to it and I will be honest and admit, although I know the basic principles of a frequency counter I did not know ALL that was going on here…particularly the decimal point positioning circuitry. However, this is an excellent unit to learn on…it is not too complex, no microprocessors or firmwate and easy to get into without extender cards. I’m so glad it eas just that 7490 second counter…still wonder how such a failure occurrs initially though. So a frequency counter can be as simple or as complex as you like. Compared to my expeirences with frequency counters, this oe is about in the middle. The basic pronciple is…get an input, Turn it into s nice digital pulse train, feed it through a gate….so, say the gate opens for one second and lets through 2738 pulses, then after the gate is the counter, and it counts to 2738 in the time the gate is open. Normally this count up is not displayed, however some counters to have settings that allow the count up to be viewed, but normally, while the counters count up, the display is either blanked or displaying the previous count the last time the gate was open, then once the new count is finished at the closing of the gate, the new reading is transferred to the display and it updates. And that’s the badic guts of a frequency counter! In this unit, the all important gate is U425A. If it is desired to count much higher frequencies, a prescaler can be placed ahaid of the gate, say this divides by 1000, so three divide by 5’s and three divide by 2 toggle F/Fs to give a nice 1:1 mark-space output. Al you need fo is move the decimal place over on the display. Sure, you win’t have 10 digit resolution, loke youd get adding more counters and displays AFTER the gate, but often thay simply can’t count fast enough anyway. I think that 4000 CMOS os godd gor about 7MHz, and Basic 7400 TTL reaches to about 20MHz. So this goes up to 250MHz, far higher than either of those logic families can reach…so how is it done? Take a look at U390 and U400, these little buggers ate ECL logic. One of yhem is an MC10131, this is a member of the 10,000 series ECL family…almost always in Cerdip packages..they sre quite power hungey, they have oddball logic levels, but they are faster than slipping in catsh!t on a hot yin roof! I’m so, so glad that boring old 7490 “shot through like a Bondi Tram” and not any of the 10k or 1.6K ECL, because, although they sre just a few toggle flip-flops, those buggers sre ExPENSIVE and relatively hard to get. If you ever see ipany old non test gear at the scrap metal yard and there are 10K or 100K series ECL in there…GRAB IT! I have a HP 5326A nicie frequency counter, it is vaguely similar to this, ECL prescalers, TTL gate and counters…but it does period, stopwatch etc., as well as frequency, so there are TONS of gates inside ot rearrange between those functions as well as the basic gate controlled by a crystal oscillator. There is lots of level conversion from ECL to TTL and back, so it is a real rat’s nest! Years ago, one NAMD gate in 10K ECL chip in the stopped NAND ing. I had no replacements of that exact gate arrangment, but I did have some other, different ECL chips that did have uncomitted NAND gates inside, so I botched this chip to the back of the board on the opposite side to the one quad NAND with the single dead gate in it….the other three NANDs in it were still working. So I connected the NAND from the different chip underneath on place of the faulty one…and it worked again. Bloody ugly fix, but it eas all I could do i the face of having no exact replacement, and the exact replacement costing fifty bucks! So, yes, nice learning curve here on a not-too-complex freqiency counter. Take a look online st the circuit disgram of the Australian sine generator, the AWA G232, it has a frequency counter in it…it only goes up to 100KHz or do, so no fancy ECL there, but see if you can identify the all important heart of it, yhe GATE, amongst the sea of TTL in its counter! cheers.
Thanks Globe Collector! Always appreciate your wealth of knowledge. I do appreciate a repair that can teach me a new concept and is a bit different than your run of the mill power supply issues - which I admit I spent too much time on. Speaking of which, I don't know if you saw my note in description, but I did end up finding out why the voltage rails were low - it was a bad LM723! Very interesting. I put it on a breadboard and built the test circuit described in its datasheet - 12V input, 5V output. And it was outputting ~5.5V. just like when it was in the SG503, it was outputting a higher voltage than it should be. Interestingly, the reference voltage was still bang on at ~7.2V. So maybe an issue with the internal error amplifier? Maybe the lower than normal voltage rails killed the 7490? The unit was operating outside the recommended minimum supply voltage for these TTLs - could that cause them to go bad? The end user would not have known without measuring the voltage rails because clearly it will still work at ~4.7V (who knows for how long...) Also, another fault is the 10MHz setting - the output will go unlevel at this setting. I determined it was an issue with the frequency setting switch assembly. This switch is a two in one rotary wafer and cam switch with those fancy gold flags and drum actuator that Tektronix loves to use. But it seems the issue was with the rotary wafer part. I found a contact for the 10MHz setting that was slightly lifted. I removed the whole switch assembly, cleaned it, and bent the contact back into place. Buzzed every setting with my meter for continuity and everything looked good. Reinstalled it and the problem persists... Now it will only go unlevel from 8-10MHz (10MHz setting is range from ~8.8 to 26MHz). So I may admit defeat on this one... I don't feel like taking the switch assembly apart again 😂
Yeah, sorry you had to explain it all here, as I had just watched your next video where you seay almost the same thing. I do wonder why that 7490 failed…maybe it was because the Vcc fell too low and sone of the transistors inside went out of saturation and in to their linear regions. Actually, Inshould try this, set up a 7400 series gate circuit…operating, then gradually dectrase the supply and see p, not only where it stops…but HOW it stops! i’ll continue my comment over under your next video, so everybody else csn see it too.
Follow Up Note: At @41:36 I check the 5.2V rail one last time and see that it is still low at ~4.7V. I speculate it could be due to the lower voltages supplied by the TM5000. This turned out to be false! It was actually a faulty LM723! I replaced it and now the voltages are in spec. I may do a short follow-up vid going over some of additional repairs (including this one) required to make this SG503 fully operational. Thanks for watching!
Nice work!
thanks Charles!
Nice repair…good to get down into the “guts” of a “sea of TTL gates” from time to time.
I was “following along on my own download of the circuit”…I did not know or pre empt the fault before you got to it and I will be honest and admit, although I know the basic principles of a frequency counter I did not know ALL that was going on here…particularly the decimal point positioning circuitry.
However, this is an excellent unit to learn on…it is not too complex, no microprocessors or firmwate and easy to get into without extender cards.
I’m so glad it eas just that 7490 second counter…still wonder how such a failure occurrs initially though.
So a frequency counter can be as simple or as complex as you like. Compared to my expeirences with frequency counters, this oe is about in the middle. The basic pronciple is…get an input, Turn it into s nice digital pulse train, feed it through a gate….so, say the gate opens for one second and lets through 2738 pulses, then after the gate is the counter, and it counts to 2738 in the time the gate is open. Normally this count up is not displayed, however some counters to have settings that allow the count up to be viewed, but normally, while the counters count up, the display is either blanked or displaying the previous count the last time the gate was open, then once the new count is finished at the closing of the gate, the new reading is transferred to the display and it updates. And that’s the badic guts of a frequency counter!
In this unit, the all important gate is U425A. If it is desired to count much higher frequencies, a prescaler can be placed ahaid of the gate, say this divides by 1000, so three divide by 5’s and three divide by 2 toggle F/Fs to give a nice 1:1 mark-space output. Al you need fo is move the decimal place over on the display. Sure, you win’t have 10 digit resolution, loke youd get adding more counters and displays AFTER the gate, but often thay simply can’t count fast enough anyway.
I think that 4000 CMOS os godd gor about 7MHz, and Basic 7400 TTL reaches to about 20MHz. So this goes up to 250MHz, far higher than either of those logic families can reach…so how is it done? Take a look at U390 and U400, these little buggers ate ECL logic. One of yhem is an MC10131, this is a member of the 10,000 series ECL family…almost always in Cerdip packages..they sre quite power hungey, they have oddball logic levels, but they are faster than slipping in catsh!t on a hot yin roof!
I’m so, so glad that boring old 7490 “shot through like a Bondi Tram” and not any of the 10k or 1.6K ECL, because, although they sre just a few toggle flip-flops, those buggers sre ExPENSIVE and relatively hard to get. If you ever see ipany old non test gear at the scrap metal yard and there are 10K or 100K series ECL in there…GRAB IT!
I have a HP 5326A nicie frequency counter, it is vaguely similar to this, ECL prescalers, TTL gate and counters…but it does period, stopwatch etc., as well as frequency, so there are TONS of gates inside ot rearrange between those functions as well as the basic gate controlled by a crystal oscillator. There is lots of level conversion from ECL to TTL and back, so it is a real rat’s nest!
Years ago, one NAMD gate in 10K ECL chip in the stopped NAND ing. I had no replacements of that exact gate arrangment, but I did have some other, different ECL chips that did have uncomitted NAND gates inside, so I botched this chip to the back of the board on the opposite side to the one quad NAND with the single dead gate in it….the other three NANDs in it were still working. So I connected the NAND from the different chip underneath on place of the faulty one…and it worked again. Bloody ugly fix, but it eas all I could do i the face of having no exact replacement, and the exact replacement costing fifty bucks!
So, yes, nice learning curve here on a not-too-complex freqiency counter. Take a look online st the circuit disgram of the Australian sine generator, the AWA G232, it has a frequency counter in it…it only goes up to 100KHz or do, so no fancy ECL there, but see if you can identify the all important heart of it, yhe GATE, amongst the sea of TTL in its counter!
cheers.
Thanks Globe Collector! Always appreciate your wealth of knowledge. I do appreciate a repair that can teach me a new concept and is a bit different than your run of the mill power supply issues - which I admit I spent too much time on.
Speaking of which, I don't know if you saw my note in description, but I did end up finding out why the voltage rails were low - it was a bad LM723! Very interesting. I put it on a breadboard and built the test circuit described in its datasheet - 12V input, 5V output. And it was outputting ~5.5V. just like when it was in the SG503, it was outputting a higher voltage than it should be. Interestingly, the reference voltage was still bang on at ~7.2V. So maybe an issue with the internal error amplifier? Maybe the lower than normal voltage rails killed the 7490? The unit was operating outside the recommended minimum supply voltage for these TTLs - could that cause them to go bad? The end user would not have known without measuring the voltage rails because clearly it will still work at ~4.7V (who knows for how long...)
Also, another fault is the 10MHz setting - the output will go unlevel at this setting. I determined it was an issue with the frequency setting switch assembly. This switch is a two in one rotary wafer and cam switch with those fancy gold flags and drum actuator that Tektronix loves to use. But it seems the issue was with the rotary wafer part. I found a contact for the 10MHz setting that was slightly lifted. I removed the whole switch assembly, cleaned it, and bent the contact back into place. Buzzed every setting with my meter for continuity and everything looked good. Reinstalled it and the problem persists... Now it will only go unlevel from 8-10MHz (10MHz setting is range from ~8.8 to 26MHz). So I may admit defeat on this one... I don't feel like taking the switch assembly apart again 😂
Yeah, sorry you had to explain it all here, as I had just watched your next video where you seay almost the same thing. I do wonder why that 7490 failed…maybe it was because the Vcc fell too low and sone of the transistors inside went out of saturation and in to their linear regions. Actually, Inshould try this, set up a 7400 series gate circuit…operating, then gradually dectrase the supply and see p, not only where it stops…but HOW it stops! i’ll continue my comment over under your next video, so everybody else csn see it too.
sacrifice a 7490 for science! 😂@@thanhhuynh272
Nice one !!!.....cheers.
Thanks Andy! I was trying to pay more attention my editing on this one!
I noticed ! great job !@@fmashockie
so they say about that L660 as possible issue...
see below...
"Display Is All Zeros With No Variations - Suspect L660 is open."
Issue was resolved by the end of the video. It was one of the TTL counters that had gone bad
Good job! Don't be so critical of yourself.
Thanks Joel! I appreciate the feedback!