Yet another butchered 250

[ocdecio]

And my only conclusion, at this point, is that the gain of the differential pair is absolutely critical in getting the bias voltage correct on the positive half of the Amplifier.

Interesting. If you are replacing SPS1928s they definitely need to have high gain, in the 700 range. The 240 uses 2N5961 from factory which has lower hFE but it uses a pair of transistors in the bias section so these things could be related.

2N4250s also have high gain, in the 400-500 area, I managed to get some and used them in the 1200 tone board. I find it interesting that the 250 reaches 125W because it has +-58V rails whereas the 1200 is rated at 100W and it has +-52V rais. The amp boards are basically the same.

 

[rickl]

Thanks guys for working through the 250. When I got mine working, I didn't have as much dialog. my power amp seems to work but I haven't done much stress testing. Plus I have other amps that get swapped in.

I think this thread will help others with broken 240/250 amps.

I'll update my thread with this thread so if someone stumbles on my thread they can find this one too.

Good job!

 

[goldswimmerb]

Thanks guys for working through the 250. When I got mine working, I didn't have as much dialog. my power amp seems to work but I haven't done much stress testing. Plus I have other amps that get swapped in.

I think this thread will help others with broken 240/250 amps.

I'll update my thread with this thread so if someone stumbles on my thread they can find this one too.

Good job!

I've actually been reading through your thread, lots of good details in there. I'm just trying to be as detailed as possible here to help others in the future.

 

[goldswimmerb]

Interesting. If you are replacing SPS1928s they definitely need to have high gain, in the 700 range. The 240 uses 2N5961 from factory which has lower hFE but it uses a pair of transistors in the bias section so these things could be related.

2N4250s also have high gain, in the 400-500 area, I managed to get some and used them in the 1200 tone board. I find it interesting that the 250 reaches 125W because it has +-58V rails whereas the 1200 is rated at 100W and it has +-52V rais. The amp boards are basically the same.

From what I found the PN4250 is supposed to be a drop on equivalent for the 2N4250.

The significant redesign on later units probably allowed for lower gain diff pairs by making up for the lack of gain by using a 2nd Transistor in the bias stage.

 

[rich l]

interestingly enough, I've been running mine on the bench with the decapitated original driver that I posted the pic of earlier, and it seems to have stabilized! It has maybe 10 hours or so of idling and a couple of hours of fairly high power audio running into dummy load and the bias settles in at around 13mV and offset +- 5mV or so so far. The right (suspect) channel seems to bounce around more but stays within range. Do you think it would be idiotic to epoxy the cap back on the driver and call it good? As long as I don't have to pull it again, the loose pins won't be an issue. Seems wrong but I think I,ve got it stable.

 

[rich l]

Thanks guys for working through the 250. When I got mine working, I didn't have as much dialog. my power amp seems to work but I haven't done much stress testing. Plus I have other amps that get swapped in.

I think this thread will help others with broken 240/250 amps.

I'll update my thread with this thread so if someone stumbles on my thread they can find this one too.

Good job!

when I jumped into repairing the 240, I had no idea it would be this challenging. My problem is, I forget what I was doing by the time parts arrive so its almost like starting over when I get back to it. The few threads on these amps have helped a lot. Thanks everyone!

 

[goldswimmerb]

interestingly enough, I've been running mine on the bench with the decapitated original driver that I posted the pic of earlier, and it seems to have stabilized! It has maybe 10 hours or so of idling and a couple of hours of fairly high power audio running into dummy load and the bias settles in at around 13mV and offset +- 5mV or so so far. The right (suspect) channel seems to bounce around more but stays within range. Do you think it would be idiotic to epoxy the cap back on the driver and call it good? As long as I don't have to pull it again, the loose pins won't be an issue. Seems wrong but I think I,ve got it stable.

I wouldn't be too worried about the fluctuations, as long as you're not seeing erratic spikes, or the bias rising with heat (indicative of thermal runaway) I would say you're fine, as long as you've verified the gain of the driver. Both of mine were reading well below the 50hfe minimum Marantz prescribed, however, my experience with the amps so far has been that the predrivers are the most critical, and for the most part, as long as the drivers are reasonably close you're fine.

As long as you're not seeing 40+mv I'd say you're good, I've seen several threads where people have run these in the 22-25mv bias range because that was where they wound up following the older biasing procedure with a watt meter.

For reference, The last 250M I did had around an 80Hfe difference between my positive and negative driver, but was completely stable, the predrivers were almost perfectly matched.

When biasing these amps I actually have to shut the AC off because any airflow in the room makes it incredibly difficult to get the bias right.

 

[rich l]

I wouldn't be too worried about the fluctuations, as long as you're not seeing erratic spikes, or the bias rising with heat (indicative of thermal runaway) I would say you're fine, as long as you've verified the gain of the driver. Both of mine were reading well below the 50hfe minimum Marantz prescribed, however, my experience with the amps so far has been that the predrivers are the most critical, and for the most part, as long as the drivers are reasonably close you're fine.

As long as you're not seeing 40+mv I'd say you're good, I've seen several threads where people have run these in the 22-25mv bias range because that was where they wound up following the older biasing procedure with a watt meter.

For reference, The last 250M I did had around an 80Hfe difference between my positive and negative driver, but was completely stable, the predrivers were almost perfectly matched.

When biasing these amps I actually have to shut the AC off because any airflow in the room makes it incredibly difficult to get the bias right.

I haven't been able to check the hfe on the drivers as it will be a real pain to remove the damaged one without the leads pulling off completely. I didn't know what Ic/Ib to use to check hfe so I swapped sides with the known good ones(both marked orange) .Thats when the leads pushed into the driver instead of the socket. What parameters are you using to check gain on these? I don't have access to a curve tracer.

 

[GlasgowGrip]

Surprisingly the TO-66 parts I wanted to try as drivers were actually even further apart than the MJEs, if going by gain rank.

@goldswimmerb: Are you saying that you are successfully using TO-220 MJE's for your drivers (Q510/Q511)? I have seen conflicting reports on AK about whether or not the TO-220 MJE parts will work for the Model 1200/250 drivers.

To be clear why I am asking, I have just rebuilt a Model 1200 with the early Mk1 power amp boards, and I used the following for drivers:
Q510 NPN = ONSemi MJE15030G
Q511 PNP = ONSemi MJE15031G

So far I have only powered up one of the amp boards, without the output transistors installed, and it seems to behave itself - I could set bias and offset. I will bite the bullet tomorrow and bring it up with the outputs installed, and hope for the best. Just reading through this thread - great thread by the way! - and I'm hoping you can confirm whether you had success with these parts. Thanks!

 

[ocdecio]

@goldswimmerb: Are you saying that you are successfully using TO-220 MJE's for your drivers (Q510/Q511)? I have seen conflicting reports on AK about whether or not the TO-220 MJE parts will work for the Model 1200/250 drivers.

To be clear why I am asking, I have just rebuilt a Model 1200 with the early Mk1 power amp boards, and I used the following for drivers:
Q510 NPN = ONSemi MJE15030G
Q511 PNP = ONSemi MJE15031G

So far I have only powered up one of the amp boards, without the output transistors installed, and it seems to behave itself - I could set bias and offset. I will bite the bullet tomorrow and bring it up with the outputs installed, and hope for the best. Just reading through this thread - great thread by the way! - and I'm hoping you can confirm whether you had success with these parts. Thanks!

How close are the MJE's hFE?

 

[goldswimmerb]

@goldswimmerb: Are you saying that you are successfully using TO-220 MJE's for your drivers (Q510/Q511)? I have seen conflicting reports on AK about whether or not the TO-220 MJE parts will work for the Model 1200/250 drivers.

To be clear why I am asking, I have just rebuilt a Model 1200 with the early Mk1 power amp boards, and I used the following for drivers:
Q510 NPN = ONSemi MJE15030G
Q511 PNP = ONSemi MJE15031G

So far I have only powered up one of the amp boards, without the output transistors installed, and it seems to behave itself - I could set bias and offset. I will bite the bullet tomorrow and bring it up with the outputs installed, and hope for the best. Just reading through this thread - great thread by the way! - and I'm hoping you can confirm whether you had success with these parts. Thanks!

Yes I am, I used MJE15031 and MJE15032 as the 32 consistently had lower gain than the 15030 making a closer match. At room temperature the HFE on my 15032 was something like 174, whereas my 15031 was in the range of 80-90 if I recall. Which would be fairly close to having them "within one color group" of each other as per the bulliten.

Keep in mind the HFE will change during operation due to heat.

From past experience with a 250M it would seem that the predriver gain is much more critical.

That being said, DO NOT test with your outputs until you are pretty confident.

I highly recommend running it without the outputs on the DBT to make sure that you have +.55v to +.8v on the can of your positive predriver and -.55 to -.8v on the negative can. That is your bias voltage, and should be adjustable with the pot. Adjust that so you have around .6v on each before putting in the outputs. If either has a significant deviation from the other you'll probably pop something again.

To add, I did aquire some 2SA483/2SC783 to test with, but I was seeing 50 to 70hfe on one and 140-170 on the other, which would have put me in the same situation as using the MJE parts.

I've been running the repaired channel of the 250 on my ELAC UB5 (4ohm speakers) and thus far have had no issues.

 

[GlasgowGrip]

Yes I am, I used MJE15031 and MJE15032 as the 32 consistently had lower gain than the 15030 making a closer match. At room temperature the HFE on my 15032 was something like 174, whereas my 15031 was in the range of 80-90 if I recall. Which would be fairly close to having them "within one color group" of each other as per the bulliten.

Thank you, this is fantastic information!

My hFE's are:
Q510 - MJE15030 = 172
Q511 - MJE15031 = 268

I do have some MJE15032's on hand, so I just got those out to test, but their hFE is only 108, so I will stick with the 15030's. Not sure why my 15031's are so much higher than yours.

The good news is my pre-drivers came out very close:
Q507 - 2N5415 = 104
Q508 - 2N3440 = 106



Anyway, I will run this board without the outputs for a while more, and make careful measurements of the pre-driver can voltages. Thanks again for the advice! - Steve

 

[goldswimmerb]

Thank you, this is fantastic information!

My hFE's are:
Q510 - MJE15030 = 172
Q511 - MJE15031 = 268

I do have some MJE15032's on hand, so I just got those out to test, but their hFE is only 108, so I will stick with the 15030's. Not sure why my 15031's are so much higher than yours.

The good news is my pre-drivers came out very close:
Q507 - 2N5415 = 104
Q508 - 2N3440 = 106



Anyway, I will run this board without the outputs for a while more, and make careful measurements of the pre-driver can voltages. Thanks again for the advice! - Steve

Went back and verified, I used MJE15032/33

My 33's are in the range of 150-200, most around 175, and my 32s are 80-120 with most being around 85-100ish. I picked what would be a high red or high yellow on the bulliten. I think yours might be too far apart. Mine weren't any farther apart than the other subsistites I had for the drivers so I figured I'd try it.

 

[rich l]

Went back and verified, I used MJE15032/33

My 33's are in the range of 150-200, most around 175, and my 32s are 80-120 with most being around 85-100ish. I picked what would be a high red or high yellow on the bulliten. I think yours might be too far apart. Mine weren't any farther apart than the other subsistites I had for the drivers so I figured I'd try it.

are you checking the drivers with a small signal transistor tester or are you checking at a higher Ic/Ib ?

 

[goldswimmerb]

are you checking the drivers with a small signal transistor tester or are you checking at a higher Ic/Ib ?

Just a small signal tester, Atlas DCA75 pro or whatever. It'll do basic curve tracing too.

 

[GlasgowGrip]

...I highly recommend running it without the outputs on the DBT to make sure that you have +.55v to +.8v on the can of your positive predriver and -.55 to -.8v on the negative can. That is your bias voltage, and should be adjustable with the pot. Adjust that so you have around .6v on each before putting in the outputs. If either has a significant deviation from the other you'll probably pop something again.

Okay, so when I run it on the DBT with no outputs installed, I have +/-0.6v on the predriver cans, and they respond to adjustment of the bias pot. It will slowly drift up or down, and if set low enough it will stay constant. Same if I measure the bias the standard way (as if the outputs were installed even though they are not). But I think I may have a problem - if I set bias to the 12-15mV range, it will not remain stable and will slowly start to run away if I don't back off the pot. Is this because the output transistors aren't there to warm up the heat sink? Or do you think it indicates a problem with the bias compensation circuit? I am re-using the original germanium diode CR801 installed on top of the upper predriver Q507 heatsink, but the original Q801 was kaput so I replaced it with an MJE180, hFE = 142. Should I just throw in the outputs and see if that adds enough heat to the bias thermal feedback loop to stabilize it? Yikes.

 

[ocdecio]

The "epoxy transistor" SPS1950 remains a mistery to me. I was comparing the bias circuit in the 1200 with the 32:



32:



The circuits are virtually the same. I suspect SPS1950 is in fact MPSA20 (or something similar with high gain); fortunately my 1200 has both in reasonable shape and they measured hFE of 400 or more which could be the issue here - I know MJE180 is the correct heatsink bias transistor but only in the 1200B and later 240/250 models which use a pair of transistors (the other being MPS6519) instead of the diode.

 

[GlasgowGrip]

The "epoxy transistor" SPS1950 remains a mistery to me. I was comparing the bias circuit in the 1200 with the 32:

The circuits are virtually the same. I suspect SPS1950 is in fact MPSA20 (or something similar with high gain); fortunately my 1200 has both in reasonable shape and they measured hFE of 400 or more which could be the issue here - I know MJE180 is the correct heatsink bias transistor but only in the 1200B and later 240/250 models which use a pair of transistors (the other being MPS6519) instead of the diode.

Ah yes... I see what you mean. Great catch on the 32 as a comp!

The one SPS1950 in my 1200 is still intact and measures hFE=410, so I agree the MJE180 seems to be too low gain. I will have to source a different sub. The MPSA20 looks like a pretty vanilla small signal transistor in a TO-92 package - hmmmm, might explain why they potted it in a block of epoxy, I think that fits with your theory. I have some other small signal NPN's I might try to sub in there just for kicks... they're all TO-92's, so will have to rig up a temporary binding to the heatsink. If it works, I may end up potting my own "epoxy transistor" for someone else to figure out 50 years from now, haha.

 

[ocdecio]

The 32 is pretty clever in which it installs both the diode and the transistor on the back of the board, bent just right to touch the heatsink. So no weird wires going everywhere. The diode that I see installed is the same 1N542 germanium. I agree, MSPS20 is a pretty common transistor, easily obtainable and it would be great to design a contraption or as you said mold it in epoxy. I will be following closely what you come up with, it may come handy when/if I need to replace mine.

 

[GlasgowGrip]

The 32 is pretty clever in which it installs both the diode and the transistor on the back of the board, bent just right to touch the heatsink. So no weird wires going everywhere. The diode that I see installed is the same 1N542 germanium. I agree, MSPS20 is a pretty common transistor, easily obtainable and it would be great to design a contraption or as you said mold it in epoxy. I will be following closely what you come up with, it may come handy when/if I need to replace mine.

The only NPN's I have on hand with high enough gain are BC547, BC550 and KSC945. None are ideal subs in terms of power rating. The BC547's and KSC945's all come in almost exactly at hFE=400, and the BC550's come in anywhere from hFE 574 - 628, so maybe a little too much gain. Until I can source a legit MPSA20, I think I will try the BC547 with hFE = 400. Power handling for BC547 is 500mw vs. 625mw for the original MPSA20, but hopefully that's plenty. BC547 also has a higher fT of 300MHz vs. 125 MHz for the MPSA20.

Unfortunately I won't be able to try out the BC547 until tonight.