2000x Phono Board & 2SC871

[ConradH]

I read a post where someone replaced the 2SC871s with KSC1845s in the phono board and was unhappy with the sound quality. They put them back and the sound quality returned. This seemed a bit inexplicable to me, as the KSC1845s work fine in the tone control board. The measured response of the phono board leaves something to be desired so I put the circuit in LTSpice to see if it could be improved. Not done yet, but here are some things I've learned-

1) Important- they switch the eq in and out of the phono board and ALL signals pass through this stage, phono or not. A problem in the phono board can affect everything.
2) There is a peak of several dB right in the record warp area, and a general rise towards the bass. The whole response has a decided tilt. Call it rolled off highs or enhanced bass. In flat mode, non-RIAA, the response is better.
3) The bass response of the circuit is transistor dependent, particularly the peak. It's probably just the hfe of the transistor, which explains the sound quality difference.
4) The input capacitors (1.5 uF special Elna CS series) often get leaky. The phono response is a bit dependent on the value of these caps.
5) The 47 uF emitter bypass cap on on the second transistor appears to be wrong by a factor of ten or so, depending on what transistors are used.

In any rebuild, the two input caps have to be replaced. Right now I'd suggest 2.2 uF low leakage types, but no higher than that. The low frequency peak can be controlled by increasing the 47 uF caps to 220 uF or even 470 uF.

One could probably make the RIAA response flatter by messing with the eq components, but IMO the signature of the 2000x is a result of the values chosen and you might not be happy with a more clinical sound.

More to follow as I implement and test this stuff.

 

[Hyperion]

Interesting Conrad, I've seen a few stories here on AK about transistor swapping causing sound quality issues, and experienced some myself.

 

[ConradH]

Learned a bit more in LTSpice, though need to confirm it in the actual hardware. It shouldn't matter too much what the input transistor is, so long as it's quiet and has decent hfe, say 150 or greater. Quite a bit greater is still OK. The KSC1845 should be good there as it's very quiet and typically has a lot of gain.

The second transistor is optimal with an hfe of about 150. Any less and you get bass rolloff. Any more and you get very low frequency peaking and a rising bass response. I'm guessing something like a BC847A (never tried one) might be good, or one of the many parts that has an hfe suffix for the lowest bin. The emitter bypass for that transistor is probably best at 220 uF instead of 47 uF.

It's not considered good practice to design circuits that are sensitive to transistor hfe, but an RIAA preamp can be tough. IMO, if you absolutely have to change the 2nd transistor, you should have an inverse RIAA network and check the response from 1 Hz on up. Some of the severe peaking can occur down at a couple Hz, depending on the hfe and caps used. I don't think the curve is specified that low, but the common inverse circuits will still show you what's happening.

I suspect Sansui just used parts with slightly higher than optimal gain, figuring a bit of bass boost was vastly better than a falling curve.

 

[larryderouin]

This will probably apply to the "A" with the same board, and parts kit. I use KSC1845's in them without problem, although I do notice now not needing the Loudness button and actually backing off Bass 1 or 2 notches before center.

 

[Leestereo]

...The second transistor is optimal with an hfe of about 150. Any less and you get bass rolloff. Any more and you get very low frequency peaking and a rising bass response. I'm guessing something like a BC847A (never tried one) might be good, or one of the many parts that has an hfe suffix for the lowest bin...

I had a similar issue with the AU-7500 phono stage, the second transistor was a 2SC1313 and the initial use of a KSC1845 replacement caused problems; the solution was to use a KSC1815 (hFE of ~200); schematic of the phono stage can be seen here.

 

[sregor]

Don't think it's the exact same, but similar - problem was corrected by changing the ksc1945 to ksc1815 in this thread. I think the 1845 lose performance when their base-collector voltage gets too low. My 2 cents

 

[ConradH]

Interesting stuff. It's actually not the rising bass that concerns me, since many people will like that. The worry is having a gain peak in the record warp area, causing a lot of woofer pumping. I don't know if it's a real world issue, especially with a cap coupled amp, but it just seems like a bad place to have a peak.

 

[spark1]

Pulling this thread out of mothballs because I'm in the middle of restoring my 2000X. I was wondering if any further work had been done by anybody as regards using something other than KSC1845's to replace the SC871s on the equalizer board...especially the second pair (not the inputs) I was thinking about using 1815s there, with 1845s for the inputs, to avoid/minimize additional bass rise.

I would also be very interested in anyone has actually tried the higher value emitter bypass caps (220uf, for example) to replace the original 47uf caps, and what you thought of the effect on the sound of the amp. I don't necessarily want to improve upon its stock performance as far as the bass is concerned, but don't want to amplify/exaggerate it either.

Thanks very much for any insights or suggestions.

 

[spark1]

To better phrase my question about using 220uf caps to replace the 47uf emitter bypass caps...would the recommended value be the same even if I ended up keeping the 871s as the second pair of transistors? And what would it be if I replace the 871s transistors with the 1815YTA?

Thank you!

 

[spark1]

OK, I'll ask a different question. To keep the 2000X sound as "original" as possible but still get rid of the noisy 2SC871s on the EQ board, what would be the best replacement for the second set of transistors (I plan on using 1845s on the input pair, and no changes to capacitor values)?

If I understand the issue correctly, and contrary to the LTSPICE-based recommendation for a much lower gain transistor (above), to keep the existing characteristics of the amp, I need something with a gain comparable to the 2SC871, which appears to be around 600 from what I can figure out.

I've seen threads which suggest the KSC1845EBU would offer higher gain, but when I look at the datasheets, it looks to me like it is not much different from the KSC1845FTA. And in any case, the EBU version has been obsoleted.

But of course I could be totally off base on any/all of my assumptions!

Thank you for any direction you can provide!

 

[dly66]

I tried the 220 uF bypass cap and it has an issue interacting with certain circuits. FM, PHONO 1 and 2 all acted normal. AM tuner output became rectified, peaking and dying out at about twice per second. AUX fed by source similar but the dips did not die out just volume got lower in the dips. Both very annoying to listen to. Putting back in a 47 uF replacement matching the old value now the AM tuner output and AUX behave as they should.

I used the KSC1845 on my 2000A. It may be a bit brighter than some of my other receivers. Also did Conrad's mods to it. The screwed up HH Scott 342-C I just restored, it was fun chasing down its issues, once in good working order has a similar sound quality to the 2000A even though it is not cap coupled and has NPN/PNP pairs at the outputs. At least with my work room KLH Model 17s. They appear to have been produced in the same time period. The Sansui may be a little brighter than the Scott. I like that sound so they are both favored in my collection, the Sansui more so. There could be some difference in sound from the caps since mostly Silmics were used in the audio path of the Sansui and PWs in the Scott in addition to the different amp design. I have not tried the Scott with the Advent Legacy IIs in the living room. Some of my smaller amps seem to run out of gas there but the 2000A seems to take it in stride more like my 45-65 wpc pieces.