transistor tester (a $20 component tester or, in my case, a Peak Atlas DCA-75 Pro)
Recapping and restoring and upgrading a Heathkit AR-1515 receiver
- Thread starter dlucy
- Start date
transistor tester (a $20 component tester or, in my case, a Peak Atlas DCA-75 Pro)
I built my AR-1515 in like in 1982. Still have it and it works great.
But over the years I have made several modifications. As usual, Heathkit had some slight blind spots in their design.
(1) The dang power supply. Huge diodes and capacitors. Which are a good thing, but when you pressed the power button the switch would put out quite a Snap!, the house lights would dim, and the power transformer would hum and thump. Something had to be done! So I hillbilly-engineered a soft-start solution-- a relay, a power resistor, and well, nothing else. The resistor limits the primary current. The relay coil voltage is picked to close the relay when the DC gets up to about 3/4. Sorry I don't recall the values I used but it should only take a little experimentation to find the optimal values. Now the capacitors charge up for about 1 second before the relay pulls in and applies full power. I'm sure the power button and other parts appreciate the soft-start.
(2) The danged FM discriminator. Heath used an innovative ECL one-shot as an FM demodulator. Fine in principle, but in implementation they provided very wishy-washy DC step-down to the ECL chip. So shaky that the voltage and therefore the center-tuning would drift as the chips warmed up. Very annoying. I don't recall exactly what I did, but it involved a 5-volt regulator bootstrapped up to provide the correct and stable Vcc. Tunable regulators didn't exist then. Nowadays I would use a LM337 or even more modern alternative.
(3) The LED display drivers run a bit on the hot side, so I epoxied some TO-5 transistor heatsinks to them. Much better.
(4) The proprietary speaker output connectors looked flimsy and I dreaded losing the unobtanium plugs and then being SOL, so I hacked up the back panel a bit and put in your basic proper 5-way banana terminals.
(5) The FM antenna input was just 300 ohms, so I added a more modern cable-TV connector and I think a 300 to 75 ohm balun. Much better.
(6) The power supply diodes lacked RF bypassing, so I added some 0.1uF capacitors across each diode. Not 100% necessary but my previous Dynaco amplifier put out a very strong RF hash that killed all of the AM band, hash from the 3-amp diodes, so I've always been wary of unbypassed diodes.
... the receiver has been running just fine for the last 39 years or so so I must have done something right.
Regards,
George
But over the years I have made several modifications. As usual, Heathkit had some slight blind spots in their design.
(1) The dang power supply. Huge diodes and capacitors. Which are a good thing, but when you pressed the power button the switch would put out quite a Snap!, the house lights would dim, and the power transformer would hum and thump. Something had to be done! So I hillbilly-engineered a soft-start solution-- a relay, a power resistor, and well, nothing else. The resistor limits the primary current. The relay coil voltage is picked to close the relay when the DC gets up to about 3/4. Sorry I don't recall the values I used but it should only take a little experimentation to find the optimal values. Now the capacitors charge up for about 1 second before the relay pulls in and applies full power. I'm sure the power button and other parts appreciate the soft-start.
(2) The danged FM discriminator. Heath used an innovative ECL one-shot as an FM demodulator. Fine in principle, but in implementation they provided very wishy-washy DC step-down to the ECL chip. So shaky that the voltage and therefore the center-tuning would drift as the chips warmed up. Very annoying. I don't recall exactly what I did, but it involved a 5-volt regulator bootstrapped up to provide the correct and stable Vcc. Tunable regulators didn't exist then. Nowadays I would use a LM337 or even more modern alternative.
(3) The LED display drivers run a bit on the hot side, so I epoxied some TO-5 transistor heatsinks to them. Much better.
(4) The proprietary speaker output connectors looked flimsy and I dreaded losing the unobtanium plugs and then being SOL, so I hacked up the back panel a bit and put in your basic proper 5-way banana terminals.
(5) The FM antenna input was just 300 ohms, so I added a more modern cable-TV connector and I think a 300 to 75 ohm balun. Much better.
(6) The power supply diodes lacked RF bypassing, so I added some 0.1uF capacitors across each diode. Not 100% necessary but my previous Dynaco amplifier put out a very strong RF hash that killed all of the AM band, hash from the 3-amp diodes, so I've always been wary of unbypassed diodes.
... the receiver has been running just fine for the last 39 years or so so I must have done something right.
Regards,
George
Last edited:
There is a guide to transistor testers here: https://audiokarma.org/forums/index.php?threads/dlucy’s-guide-to-transistor-testers-and-curve-tracers.916897/
The DROK transistor tester you're talking about is fine. It will tell you if a transistor is _fundamentally_ working, it will confirm its pinouts, and it will tell you the basic hfe. Those cheap-but-work-fine testers are covered in the above link (with pointers to the manual and upgrades) and there is a great thread just about those “cheap $20 transistor testers” here: https://audiokarma.org/forums/index.php?threads/cheap-component-tester.749729/
lrb1951
JBL L110 Nut
I read through your guide, Doug and found roger2’s addition on the DMM hFE feature. My Harbor Freight DDM has this feature so according to roger2 it’s all I need to use to match transistors. Just need to make up some alligator/wire combos that I can insert into the DMM slots. No real need to spend more money if I don’t really have to. Thanks for your help, Doug.
Larry
Larry
Last edited:
lrb1951
JBL L110 Nut
George, I built my AR-1515 in 1977, so it’s some 44 years old. Mine never “dimmed the lights” when it was powered up. Even those with new 12,000 uF power supply capacitors don’t dim the lights when they are turned on. Doug mentioned in one of his earlier posts about checking out the power switch but he never elaborated on the subject. The original power switch in my AR-1515 would not work properly just about from the get go. I didn’t bother to replace it since I had bought an Audio Control Center AD-1307 with the AR-1515. I left the receiver on, plugged in into a switched outlet on the AD-1307 and used it’s power switch. The power switch Heath used in the AR-1515 could have been a bit more robust. I did buy some replacement switches from Mouser, which are very close to the original, that I have used.
I did discover that the IC readout drivers ran quite warm and did install heat sinks as you did.
Larry
I did discover that the IC readout drivers ran quite warm and did install heat sinks as you did.
Larry
Doug G.
Lunatic Member
After 45 years, the power switch on my AA-29 I built in 1976, became interrmittant. A replacement is fairly expensive so, before buying a new one, since I have repaired similar switches in the past, I removed it, disassembled it, and fixed it. It's a push-push switch and there are swinging contacts which contact fixed contacts on the switch body. The swinging contacts are operated by a kind of wedge affair connected to the push shaft.
Anyway, the originally polished silver contacts were all pitted and rough so I cleaned, smoothed, and polished them and reassembled which was the trickiest part, trying to keep everything in place while getting the cover back onto the switch body.
It works again. Whether or not it will last another 45 years is something I will probably never know.
Doug
Anyway, the originally polished silver contacts were all pitted and rough so I cleaned, smoothed, and polished them and reassembled which was the trickiest part, trying to keep everything in place while getting the cover back onto the switch body.
It works again. Whether or not it will last another 45 years is something I will probably never know.
Doug
multi-volti
New Member
Hello -
One question & one answer.
My Q: Can anyone please confirm for me whether the Heathkit AR-1515's power transformer p/n is 54-916? I have a couple but can't find a manual anywhere online to verify that. I found a list saying 54-916 is for AR-1515 but the specs made no sense.
My A: The question(s) about why would different transistors run hotter than others if they run the same voltage and current: If comparing the same package (TO-3 in this) case, and they actually DO run the same voltage and current, not all transistors in TO-3 package have the same power rating or thermal resistance from chip/die to case. But that is probably not a huge factor. The issue I think is more likely is that they may not actually run the same current, depending on the circuit. hFE may have been the same on the data sheets for the two part number transistors, but that's listed at a few operating points, not all possible conditions. Regardless of whether one part or the other had better or worse specs, if they weren't identical, they could conceivably have some different behavior at certain operating conditions.
I once repaired a 500 VA AC source that had badly melted down. Replacing it was never the first option...having someone fix it first was cheaper. It was basically an ac oscillator and power amp used for a power source at a frequency different from 50 or 60 Hz. I could not find the documentation for that exact model or replacement original p/n transistors of the same brand or vintage, so I did select modern replacements...I think they might might have been MJ21194's. Interestingly, the specifications for the new and old transistors were pretty much identical, as were the suggested applications, except the modern ones had lower leakage and one other spec I forgot was improved. I used a service manual for a larger model, but it had significant differences from the one being repaired. Both models did use the same preamp.
The repair was successful but I was concerned about how hot the thing ran even with no load. No longer supported by the manufacturer, the question of how the transistor bias was supposed to be set was not answerable. There had actually not ever been a service manual for that specific model. The larger model had some technical data, but there were some components that were factory selected.
The circuit description said the preamp input stage was Class A. I concluded (lacking any other theory) the amplifier was possibly also Class A. There was no numeric info for the bias level, and no voltages on the schematic other than the power supply VCC. I assumed that the bias was one of the parameters set up with factory-select components. It was not possible to determine how hot 'normal' was, anecdotally or quantitatively. Even the metrology lab had no idea, as that was not something they had a spec for!
I imagine anyone substituting transistors in this amp might have verified the bias was reasonably close to where it historically had been.
One question & one answer.
My Q: Can anyone please confirm for me whether the Heathkit AR-1515's power transformer p/n is 54-916? I have a couple but can't find a manual anywhere online to verify that. I found a list saying 54-916 is for AR-1515 but the specs made no sense.
My A: The question(s) about why would different transistors run hotter than others if they run the same voltage and current: If comparing the same package (TO-3 in this) case, and they actually DO run the same voltage and current, not all transistors in TO-3 package have the same power rating or thermal resistance from chip/die to case. But that is probably not a huge factor. The issue I think is more likely is that they may not actually run the same current, depending on the circuit. hFE may have been the same on the data sheets for the two part number transistors, but that's listed at a few operating points, not all possible conditions. Regardless of whether one part or the other had better or worse specs, if they weren't identical, they could conceivably have some different behavior at certain operating conditions.
I once repaired a 500 VA AC source that had badly melted down. Replacing it was never the first option...having someone fix it first was cheaper. It was basically an ac oscillator and power amp used for a power source at a frequency different from 50 or 60 Hz. I could not find the documentation for that exact model or replacement original p/n transistors of the same brand or vintage, so I did select modern replacements...I think they might might have been MJ21194's. Interestingly, the specifications for the new and old transistors were pretty much identical, as were the suggested applications, except the modern ones had lower leakage and one other spec I forgot was improved. I used a service manual for a larger model, but it had significant differences from the one being repaired. Both models did use the same preamp.
The repair was successful but I was concerned about how hot the thing ran even with no load. No longer supported by the manufacturer, the question of how the transistor bias was supposed to be set was not answerable. There had actually not ever been a service manual for that specific model. The larger model had some technical data, but there were some components that were factory selected.
The circuit description said the preamp input stage was Class A. I concluded (lacking any other theory) the amplifier was possibly also Class A. There was no numeric info for the bias level, and no voltages on the schematic other than the power supply VCC. I assumed that the bias was one of the parameters set up with factory-select components. It was not possible to determine how hot 'normal' was, anecdotally or quantitatively. Even the metrology lab had no idea, as that was not something they had a spec for!
I imagine anyone substituting transistors in this amp might have verified the bias was reasonably close to where it historically had been.
Last edited:
I wouldn't worry if the transistors run "warm", like 120 degrees F. Anything above that and I might adjust the bias. It should be easy to adjust, usually a small potentiometer going between two transistor base leads. Usually there is a small resistor of up to one ohm in the emitters or collectors of the power transistors, you can monitor the idle current across them. Figure out what the current would be for like 10 watts idle. For instance if the power is plus and minus 100 volts, 10 watts would be 50 milliamps. Across 1 ohm that would appear as 50 millivolts. Just as a wild guess I'd say 10 Watts would be plenty of idle dissipation.
One possibility if the transistors are unusually hot is oscillation. You see newer replacement transistors have much higher frequency response, into the tens of megacycles. If you replaced ancient transistors, the poor frequency response of the old ones may have been preventing oscillations in the low megahertz. A scope on the output might show a very fuzzy line, showing oscillations. The amplifier might ned some ferrite beads on the base leads.
One possibility if the transistors are unusually hot is oscillation. You see newer replacement transistors have much higher frequency response, into the tens of megacycles. If you replaced ancient transistors, the poor frequency response of the old ones may have been preventing oscillations in the low megahertz. A scope on the output might show a very fuzzy line, showing oscillations. The amplifier might ned some ferrite beads on the base leads.
Yep, I noticed the drift too, circa 1980. . I fixed it by regulating the voltage to the ECL FM discriminator chip. Heath was kinda notorious for often leaving in one little spot of poor engineering. In that area, they dropped the B+ down to very roughly 5.5 volts by using a Zener diode in series. Long ago I learned in science class, if you subtract a large raction of a large number, you're left with extra large error. Same problem here. Any drift in the B+ or Zener voltage is going to adversely affect the "5.5" volt source for the ECL chip. I fixed that by using a 7805 with a 1N4007 in its ground lead, to bring the regulated voltage up to 5.6 or so. Works perfectly. If I was doing it today I'd use an adjustable regulator like a LM337. Progress!
lrb1951
JBL L110 Nut
I fired up that over heating AA-1515 the other day and had been playing it for a few hours. I happened to touch the left heat sink and it was very hot to the touch. The right one was quite cool in comparison. I pulled the board replaced all culprit capacitors that could cause the problem, checked for cold solder joints and doubled checked the one resistor that could be a problem. Put it back together and fired it up. The left heat sink started warming up more rapidly than the right one. Shut it down again. So the problem isn’t high frequency oscillation. I removed the original output transistors and replaced them with another pair from a good board. So today I’ve had it playing since before 10 a.m. this morning and it’s now after 3 p.m. and temperature of the left heat sink has remained within 3 or 4 degrees of the right one.
I suspect the 2 that were getting hot weren’t matched or one or both we’re going out of spec.
Larry
I suspect the 2 that were getting hot weren’t matched or one or both we’re going out of spec.
Larry
Last edited:
lrb1951
JBL L110 Nut
Doug, I’m with you on building a Heathkit again. The last couple of years I have enjoyed recapping all my Heathkits. It’s not quite the same as building a new one but it’s still very enjoyable. I keep looking on eBay for another AA-1515 to recap that is on the inexpensive side. Not that I really need one. I have three that need the preamp and control boards recapped and have thought about replacing all the resistors with metal film resistors on the output amp boards, but haven’t purchased the parts yet.
Larry
Larry
Last edited:
lrb1951
JBL L110 Nut
Fred Soop, would you please elaborate on why the MJ21196G is better than the MJ802G?
I am contemplating buying new output transistors for several of my AR-1515 and AA-1515s that run on the warmer side. I suspect the problem with the ones that run warmer is that the output transistors are not closely matched pairs.
Larry
I am contemplating buying new output transistors for several of my AR-1515 and AA-1515s that run on the warmer side. I suspect the problem with the ones that run warmer is that the output transistors are not closely matched pairs.
Larry
Last edited:
fred soop
Super Member
The major difference with the MJ21196 vs MJ802 and other earlier products is much less beta droop at higher currents. Higher beta droop means that there will be more current required from the drivers at high current loads and if drive requirements approach or exceed what the VAS can supply, distortion will increase.