I am finally getting around to fixing one of my Ampzilla amps (not the Ampzilla II or Son...the original). I had the -75V supply wire that runs to the edge of the board wear away and short itself out against a sharper edge of the output board solder connections. Strange. I guess it was close for years and vibration from the fan caused it? No clue. From the short, it blew up outputs and diodes on one channel. Got it fixed, but used parts on hand that were not the right transistors (used 2N5883 & 2N5885). They are lower voltage devices, so just used my variac to bring it up enough to make sure everything was stable and working.
Now I need outputs. Did some trading with another member here and got a set of original GAS101 & GAS102 outputs. Took the amp apart today and found out someone used NTE parts on the other channel. Question, if I can't get another set of GAS101 & GAS102 for the NTE channel, should I just order MJ15022 & 23 devices? If so, how closely matched are they. I really don't feel like buying a bunch at $7 each to match. Are they close enough for me to be happy and the amp to be stable if I order what I need?
Open to suggestions. Thanks!
Now I need outputs. Did some trading with another member here and got a set of original GAS101 & GAS102 outputs. Took the amp apart today and found out someone used NTE parts on the other channel. Question, if I can't get another set of GAS101 & GAS102 for the NTE channel, should I just order MJ15022 & 23 devices? If so, how closely matched are they. I really don't feel like buying a bunch at $7 each to match. Are they close enough for me to be happy and the amp to be stable if I order what I need?
Open to suggestions. Thanks!
The outputs are series connected not parallel, so you can almost get away with 80 volt transistors. Dynamic balance during transients is the only reason you can’t. Being series instead of parallel they need not match at all to be safe, but matching hfe between NPN and PNP produces the least distortion in any circuit.
If the NTEs have never given you any grief you could just leave them if the are NTE60/61. If they worked well for any length of time they are real M15003/4. NTE180/181 although “higher” performing are more fragile and I wouldn’t use them (MJ802/4502). If you want new ON parts stick with MJ15003/4 - they have more than enough SOA for a series connected stage and higher gain at 15 amps than MJ15022/3/4/5 or MJ21193/4/5/6. If you have original Moto 2N5630/6030 on hand use them because they were the original types. Fairchild made them too, and my Dyna 400 has those.
If the NTEs have never given you any grief you could just leave them if the are NTE60/61. If they worked well for any length of time they are real M15003/4. NTE180/181 although “higher” performing are more fragile and I wouldn’t use them (MJ802/4502). If you want new ON parts stick with MJ15003/4 - they have more than enough SOA for a series connected stage and higher gain at 15 amps than MJ15022/3/4/5 or MJ21193/4/5/6. If you have original Moto 2N5630/6030 on hand use them because they were the original types. Fairchild made them too, and my Dyna 400 has those.
I have one channel's worth of 2N5630 / 6030 that are good. Got them from another member here in trade. I messaged him to see if he has another set for the other channel.
I was looking at the MJ15003 / 4, MJ15022 / 23 and the MJ21193 / 94. Looking at the data sheets, I do see some differences, but need feedback like yours to help with the choice. There is talk about the MJ21193 with good distortion characteristics and the MJ15003 says for low distortion complimentary designs. All are priced similarly at Mouser.
I appreciate the feedback.
MJ21193/4 have the best gain profile and matching, below 10 amps. Above 10 amps the gain falls off faster than MJ15003/4 though. The latter are 20A transistors. Lowest gain of the bunch is 15022/3 but they are the most rugged overall. Data sheet says 21193 has higher SOA but the jury is out on that in the real world. The 21193 does better at elevated voltage, but the old 15024 can do more than the data sheet says, below 80 volts. The “best” one really depends on what you are using them in.
Series connected - choose highest gain above 15A - 15003. (since it has to conduct all the current)
Parallel connected below 80 volt rail - choose most rugged -15024.
Parallel connected above 80 volt rail - choose highest S/B breakpoint - 21194.
Designing for best linearity, short of using modern flatpacks - choose 21194 and run parallel (Or series-parallel).
Series connected - choose highest gain above 15A - 15003. (since it has to conduct all the current)
Parallel connected below 80 volt rail - choose most rugged -15024.
Parallel connected above 80 volt rail - choose highest S/B breakpoint - 21194.
Designing for best linearity, short of using modern flatpacks - choose 21194 and run parallel (Or series-parallel).
Thanks for the detailed feedback. I am a great technician (well I think so), but not an engineer. Not sure your background @wg_ski , but I do appreciate the details.
Realizing I can't afford to buy 50 of each to match them, how critical do you think it is? I plan to get a few extra just to be safe, but not an abundance.
Realizing I can't afford to buy 50 of each to match them, how critical do you think it is? I plan to get a few extra just to be safe, but not an abundance.
IRRC wg-ski stated that the MJ15003/4 pair were developed from the original 2N5630/6030 pair. I agree wg-ski is a great wealth of knowledge, I enjoy reading and learning from his posts.
BTW I'm a GAS, SAE, and SUMO nut. I have a few Ampzilla chimneys in for rebuilds now and will be using MJ15003/4Gs.
Craig
BTW I'm a GAS, SAE, and SUMO nut. I have a few Ampzilla chimneys in for rebuilds now and will be using MJ15003/4Gs.
Craig
Before James Bongiorno passed away he was using the MJ21193/4G pair in his update program for all his babies. He did keep the original SanKen outputs if the amp came with them.
When you rebuild the Ampzilla chimneys do you make any changes on the originals or Sanken versions?
I usually don't do any circuit modifications unless there's a Service Bulletin. Sankens came in the Ampzilla II(a) and I do have some James Bongiorno tweeks for the IIa circuit. If it's a complete strip and start over I use metal film resistors in place of the carbon films, replace the electrolytics, and use modern available transistors. The early Ampzillas used MPSUxx transistors in several places and I have found that the leads break very easily on these so they get replaced with MPSWxx versions. I just rebuilt a Thaedra preamp and several of those MPSUs were already broken when it came in. My idea, right or wrong, is to build it like GAS would build it today and so far I've had good luck.
One factory mod for Ampzilla I is to the VI limiter circuit, remove L1, L2, C16, C18, R29, and R30. These components may be numbered differently depending on which version of board you have. I think there were at least three versions of the original Ampzilla board, same circuit just different parts layout.
One factory mod for Ampzilla I is to the VI limiter circuit, remove L1, L2, C16, C18, R29, and R30. These components may be numbered differently depending on which version of board you have. I think there were at least three versions of the original Ampzilla board, same circuit just different parts layout.
Last edited:
MPSW’s aren’t any more available these days than MPSUs are. Modern types work well there, and ST’s MJE340/350 work “well enough” for MPSU10/60. For the 06/56 Fairchild/on had a pair of giant TO-92s - not sure if they are still around. If you get desperate Diodes Inc/Zetex has many TO-92-like pairs that will do 1W at 25C ambient (and in general are pretty good as far as medium power higher current devices go). And the leads wont fatigue and break off after a couple decades.
I'm a bit of an Ampzilla nut. I have a few. All work but this unit. Thanks for your feedback.
I don't plan on modifying the original circuit. I do know I have another Ampzilla with the different style of boards.
Thanks for the info on this too.
I spent some time today looking into this unit. I have the chimney out and it is just a cluster of repairs over the years. I have owned it for 25 years and I know one repair was mine. However, I'm at a tipping point. 1) Do I gut this thing and replace the resistors and caps (including all the ceramic caps) with modern day stuff? The driver and VA transistors are also a mess of some of this and some of that. 2) Leave as much old stuff that doesn't need replaced and just get it working?
Q1 through Q6, plus Q8 look to be original on both channels. Q7/Q9 through Q15 are a total mess with random devices to make the amp work. I'm not concerned some are not TO-66 devices; more concerned it looks like whatever was on hand was used to fix the amp.
I repaired a SON of Ampzilla years ago and exchanged emails with James Bongiorno directly and bought some outputs from him at the time. I have some leftover parts, but not sure what I have yet. I'll probably be asking about more replacements for the driver transistors once I evaluate some more.
Q1 through Q6, plus Q8 look to be original on both channels. Q7/Q9 through Q15 are a total mess with random devices to make the amp work. I'm not concerned some are not TO-66 devices; more concerned it looks like whatever was on hand was used to fix the amp.
I repaired a SON of Ampzilla years ago and exchanged emails with James Bongiorno directly and bought some outputs from him at the time. I have some leftover parts, but not sure what I have yet. I'll probably be asking about more replacements for the driver transistors once I evaluate some more.
In the Ampzilla I just finished I used MPSW06/56 for the input diffs, the first stage of the VAS, and the regulator. Then MJE15034/35 for the second stage of the VAS and the VI limiters. Drivers are MJE15028/29. Then MJ21195/96 for outputs as I have lots of them and everything else is back ordered. Waiting on MJ15003Gs now.
As far as the aluminum TO-66s go every time I reuse aluminum cased transistors including TO-3s I end up screwing myself as the amp always seems to fail shortly after the customer gets it back so no more. I think wg_ski had something to say about aluminum transistors. The Fairchild 2N6284/87s in SAE equipment were the biggest offenders. The 2N6287s are obsolete now and hard to find so that's why I reused them. Fortunately I found a stash of them and got enough for my lifetime.
This Ampzilla will be used with another Ampzilla to drive double Dahlquist DQ-10s, a hot setup way back when.
Fun, fun, fun!!!
As far as the aluminum TO-66s go every time I reuse aluminum cased transistors including TO-3s I end up screwing myself as the amp always seems to fail shortly after the customer gets it back so no more. I think wg_ski had something to say about aluminum transistors. The Fairchild 2N6284/87s in SAE equipment were the biggest offenders. The 2N6287s are obsolete now and hard to find so that's why I reused them. Fortunately I found a stash of them and got enough for my lifetime.
This Ampzilla will be used with another Ampzilla to drive double Dahlquist DQ-10s, a hot setup way back when.
Fun, fun, fun!!!
Aluminum cased power transistors (and 3 terminal voltage regulators) have a well documented wear-out mechanism of about 5000 thermal cycles. When using a used transistor you have to ask yourself how many of them they’ve been through already.
The TO-66 VAS and drivers are nothing special - MJE1502x/3x beats them in every way. Oh yeah, there is one thing special about them - the prices that get charged. The composite gain of the MJE15029 and MJ21194 is high enough to keep the VAS operating class A and the voltage between the two outputs reasonably well balanced, but the original TO-66’s are better off if MJ15003 output are used. If there is less gain at 1A in the driver, it has to be made up with more gain at 15 in the output. 15 amp peaks are simply a fact of life with a 4 ohm load and 75V supply.
The TO-66 VAS and drivers are nothing special - MJE1502x/3x beats them in every way. Oh yeah, there is one thing special about them - the prices that get charged. The composite gain of the MJE15029 and MJ21194 is high enough to keep the VAS operating class A and the voltage between the two outputs reasonably well balanced, but the original TO-66’s are better off if MJ15003 output are used. If there is less gain at 1A in the driver, it has to be made up with more gain at 15 in the output. 15 amp peaks are simply a fact of life with a 4 ohm load and 75V supply.
Back to James' rebuild/upgrade program. All pF values use Silvered Mica and the uF values (.01, .1, etc.) use Panasonic stacked film, ECQ line? The ECQs are next to impossible nowadays, but there are other stacked films out there. If you don't want to spring for the $ilvered Mica, TDK MLCC C0G(NPO) work fine too. You could probably use these for the larger values also. There are some people than think the C0G(NP0) are a better choice than the SMs.
I keep going back to James' recommendations as he knew these amps better than anyone on the planet.
I keep going back to James' recommendations as he knew these amps better than anyone on the planet.
Status update. Two full amps rebuilt. I tested each board assembly with my variac and test output transistors to make sure DC offset was ok and sine wave looked ok on a scope. One amp has original devices for the differential pairs. The other amp was a complete mess and got all new matched 5401s & 5551s.
Will be assembling the chimneys in the amps shortly and testing further.
Will be assembling the chimneys in the amps shortly and testing further.
