I made the following circuit in LTSpice to play around with while I'm away from the amp on holiday travel. Apologies for not correctly labeling the resistors yet. 
If I input the stock resistor values for R105, R115, and R117 I get ~18mV at the output of the driver. I'm not modeling with the correct transistors so I don't expect the results to match perfectly.
Changing R105 to 68k results in 2mV offset at the output. I'm guessing that since no real load is attached to the output that my exact numbers aren't perfect but, it lends credence to the theory that tweaking R105 is the correct path.
I noticed that unbalancing the current mirror emitter degeneration resistors could also change the dc offset at the output. I'm assuming it's better to run the pair as balanced as possible and fix the problem by tweaking R105?
If I input the stock resistor values for R105, R115, and R117 I get ~18mV at the output of the driver. I'm not modeling with the correct transistors so I don't expect the results to match perfectly.
Changing R105 to 68k results in 2mV offset at the output. I'm guessing that since no real load is attached to the output that my exact numbers aren't perfect but, it lends credence to the theory that tweaking R105 is the correct path.
I noticed that unbalancing the current mirror emitter degeneration resistors could also change the dc offset at the output. I'm assuming it's better to run the pair as balanced as possible and fix the problem by tweaking R105?
No on both counts.
I'm guessing the spec called for a 100k input impedance in order not to warp the log pot characteristic around the -6 dB point too much (where the output impedance of a 100k pot reached 25 kOhms). That's, uhh, the 3 o'clock position or thereabouts? Chances are you're typically listening below -20 dB, probably more like -40 dB or even below.
Erm, this is not going to work with the VAS transistor running on thin air. You'd have to give it a current source to do the job of TR107 at the very least. Looks to be about 10-13 mA or something. BC550 would be a closer match to 2SC1845, too.I made the following circuit in LTSpice to play around with while I'm away from the amp on holiday travel. Apologies for not correctly labeling the resistors yet.
Last edited:
Of course they do - they eliminate it. These emitter followers would now have to work a bit harder (and frequency response would drop off a bit earlier in the bass), but considering they're running at a decent amount of current to begin with and should only ever see very modest levels over a few hundred mV there, I'm not overly worried. (Simulation says expect <-90 dB of dominant 2nd harmonic.)
What I am slightly worried about is the amount of series resistance stacking up between the volume pot and power amplifier input - 2k2 here, 5k6 there, another 2k2. This amp should have about 40 dB of gain, so their noise would get amplified a fair bit, depending on input stage current noise. Calculation says it may still be quite tolerable though, as in 200-300 µV of output noise, which would generally be audible but not overly bothersome with normal-sensitivity speakers. Vintage amps can be a lot worse.
As for the theory I came across that the unit was built by Technics - there isn't a single Matsushita IC in there, but loads of Hitachis. So if in doubt I'd say Hitachi, who also happened to OEM a fair few of these store brand systems around the time. Them or maybe Sanyo.
What I am slightly worried about is the amount of series resistance stacking up between the volume pot and power amplifier input - 2k2 here, 5k6 there, another 2k2. This amp should have about 40 dB of gain, so their noise would get amplified a fair bit, depending on input stage current noise. Calculation says it may still be quite tolerable though, as in 200-300 µV of output noise, which would generally be audible but not overly bothersome with normal-sensitivity speakers. Vintage amps can be a lot worse.
As for the theory I came across that the unit was built by Technics - there isn't a single Matsushita IC in there, but loads of Hitachis. So if in doubt I'd say Hitachi, who also happened to OEM a fair few of these store brand systems around the time. Them or maybe Sanyo.
I have been told that this is an NEC built unit. There's a pile of NEC transistors inside that I've not seen used in anything but their own designs, including the main output transistors.
There were two primary OEMs making receivers for MCS, NEC and Alvin, a much smaller Taiwanese outfit. The Japanese built ones seem to all come from NEC.
There were two primary OEMs making receivers for MCS, NEC and Alvin, a much smaller Taiwanese outfit. The Japanese built ones seem to all come from NEC.
Last edited:
As far as resistor noise stacking up, the good news is that at least for the AUX input, what you've seen is it. It connects directly to the volume/balance circuit shown. It's actually surprisingly quiet, I've cranked it to full volume with no input to test and there's just a small amount of noise even at that volume.
It is obvious that lots of attention was paid to getting noise down on the power supply side, so that probably shows, next to a moderate amount of gain obviously. Actually, if you used nice low ESR electrolytics to replace the critical ones, chances are that supplies are quieter than stock.
Not bad for a little store brand receiver. We've had some big name-brand components here that needed some major voltage regulator and ground routing tweaks first (*cough* Sansui AU-5900 *cough*).
Not bad for a little store brand receiver. We've had some big name-brand components here that needed some major voltage regulator and ground routing tweaks first (*cough* Sansui AU-5900 *cough*).
I've been impressed with the overall build quality and thoughtfulness of the design. The boards are really easy to work on and well laid out. The only area I see where they really scrimped is the sheet metal on the rear of the chassis (pretty thin) and the pots and switches (meh).
I replaced the main filter caps with 10,000uF high ripple capability EPCOS and all the other non-signal path 'lytics with Nichicon UPW. Signal path are Kemet films and either UKL or UES 'lytics where necessary. I'm shy rebuilding the phono stage with tighter tolerance parts and some tweaks to values to tighten the RIAA conformance.
I'll use the balance/volume control circuit as practice for doing frequency response measurements in LTSpice. I bought this receiver for $17 because I knew that some of these MCS models were half decent and I wanted something to work on that no one would cry over (except me) if I borked it up. It's turned out to be an even bigger learning experience than I expected. Modeling the circuit in spice has been super fun.
Would the lower input impedance have any effect on the balance control? It is immediately prior to the amp input.
I replaced the main filter caps with 10,000uF high ripple capability EPCOS and all the other non-signal path 'lytics with Nichicon UPW. Signal path are Kemet films and either UKL or UES 'lytics where necessary. I'm shy rebuilding the phono stage with tighter tolerance parts and some tweaks to values to tighten the RIAA conformance.
I'll use the balance/volume control circuit as practice for doing frequency response measurements in LTSpice. I bought this receiver for $17 because I knew that some of these MCS models were half decent and I wanted something to work on that no one would cry over (except me) if I borked it up. It's turned out to be an even bigger learning experience than I expected. Modeling the circuit in spice has been super fun.
Would the lower input impedance have any effect on the balance control? It is immediately prior to the amp input.
Last edited:
I've modeled the volume/balance control, please point out any errors you see. I haven't gotten the loudness circuit working yet, I need to figure out how to setup the tapped volume pot to work right.
I modeled both channels so I can change values on one and observe the changes. From a frequency response standpoint, changing to 68k instead of 100k is "near enough as to make no difference". A slight loss in output voltage but frequency response and phase remain nearly unaffected.
I modeled both channels so I can change values on one and observe the changes. From a frequency response standpoint, changing to 68k instead of 100k is "near enough as to make no difference". A slight loss in output voltage but frequency response and phase remain nearly unaffected.
Do you mean in the model? Right-click it, there should be an adjustment parameter for the tap somewhere, and make sure the pot is in the right way. You'll probably have to measure where the tap is on the real deal, and then mathematically "unparallel" (220k + 12k).
Rp = R1 * R2 / (R1 + R2) <=>
R1 * R2 = Rp * (R1 + R2)
R1 ( R2 - Rp) = R2 * Rp
R1 = R2 * Rp / (R2 - Rp)
Ta-Dah.
I'd suggest doing some .TRANsient sims. Insert these SPICE lines:
.option plotwinsize 0
.param freq=1k
.param period=1/freq ncycles=25 startrec=5
.param validcycles=ncycles-startrec
plus
.tran 0 {ncycles*period} {startrec*period} {pi/3.14*period/2e3}
.four {freq} 19 {validcycles-1} V(out-amp)
and give V3 an additional
SINE(0 283m {freq})
Note: .FOUR results (seen in error log) can be quite good if no capacitor phase shift is occurring, otherwise they may read high. If in doubt, consult the FFT. (Select plot window - View --> FFT - down to 65536 - Windows Function: Kaiser-Bessel, Beta = 20. Make sure you turn on the 'Ctrl+G'rid.)
Wow, thanks!
What I meant was that I need to figure out electrically how the pot is setup in real life. Where the tap is tapped in on the pot so I can replicate that functionality in the model.
It'll take me a bit more time to digest everything in that post.
In other news I accidentally deleted the original amp model while doing some folder cleanup.
What I meant was that I need to figure out electrically how the pot is setup in real life. Where the tap is tapped in on the pot so I can replicate that functionality in the model.
It'll take me a bit more time to digest everything in that post.
In other news I accidentally deleted the original amp model while doing some folder cleanup.
I would tweak R8 instead of R4 because R8 has no AC affects.
Also, R9 is suspect. What happens when you remove it? R9 requires current offset: Ic(Q2)>Ic(Q3,Q5,Q4) which will produce a small positive offset :Ib(Q2) > Ib(Q3). Higher gain transistors reduce the offset by reducing both Ib(Q2) and Ib(Q3).
I like the use of simulation. Play with the simulation until you have an attractive solution.
Also, R9 is suspect. What happens when you remove it? R9 requires current offset: Ic(Q2)>Ic(Q3,Q5,Q4) which will produce a small positive offset :Ib(Q2) > Ib(Q3). Higher gain transistors reduce the offset by reducing both Ib(Q2) and Ib(Q3).
I like the use of simulation. Play with the simulation until you have an attractive solution.
I really need to start numbering my simulations to match the schematic, I'm getting confused trying to keep track of things.
Edit-which R4 were you referring to? The first schematic or the second one? R17 in the second LTSpice screen grab = R4 in the first. That may be where we got mixed up. Again, I need to clean up my modeling.
Edit-which R4 were you referring to? The first schematic or the second one? R17 in the second LTSpice screen grab = R4 in the first. That may be where we got mixed up. Again, I need to clean up my modeling.
Last edited:
Going back to the numbering scheme on the schematic:
I believe R135 exists to complete TR105's base current "circuit" outside of the signal path. Without it, what would happen to that current?
It may also have some stabilization effects that I'm not knowledgeable enough to question.
I believe R135 exists to complete TR105's base current "circuit" outside of the signal path. Without it, what would happen to that current?
It may also have some stabilization effects that I'm not knowledgeable enough to question.
Last edited:
That's part of why I prefer the Linus Torvalds approach to backups for things like that.In other news I accidentally deleted the original amp model while doing some folder cleanup.
R135 is mostly just limiting current source output impedance. I can't imagine it's doing an awful lot to begin with, I mean, TR105's input impedance should be relatively low. I looked for similar circuits in the collection, didn't turn up that much, but found the Pioneer SA-610 - no such resistor in there.
At 68k ohms it's contribution to unbalancing the diff pair has to be quite small. I was getting 2mV with the input resistances matched which seems perfectly acceptable. Unless it absolutely needs to go I think I prefer to tread lightly with modifications to the original circuit. I do want to treat the offset though.
I'll get the rest of the circuit modeled into my new simulation and report back.
I'll get the rest of the circuit modeled into my new simulation and report back.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.