noise from Teac A-4010s

[js22]

So at the end of a tape I was listening to I noticed that there was a low rumbling/scratchy noise. The tape was a poorly produced from-vinyl party tape from the early 70‘s, so I assumed someone had recorded 15 minutes of a record player scratching along on the inner spiral. But no, even with the tape stopped and removed, the deck still makes the noise.

I confirmed the origin of the noise by powering the deck off and on. It's definitely the deck. I'd describe it as a light background hiss with a bunch of low frequency rumble and popping on top of that. Similar to wind on a microphone.

I've heard about leaky transistors causing noise. So that's one option. Then there's old electrolytic caps in the audio path. But do they generate noise or merely distort what passes through them?

Has anyone replaced transistors on this unit? (I don't recall OTTOMH, but it seems like these were germanium, which sounds like it could be a pain, with setting new bias levels for Si substitutes... Worth it?

 

[js22]

After posting I did some more searching. Firstly, no this isn't an Akai .

Second, I ran the RCAs directly from the preamp to the stereo amp (bypassing the recording amp in the A-4010s) and no change.

Thirdmore, while doing all this I kept listening to the noise. It definitely drifts in and out. I could see listening to this instead of that "Ocean Sounds" CD I picked up. Best part: no F'n seagulls.

 

[rojoknox]

Greetings from RojoLand!

Recapping the playback amp card is worthwhile. The transistors are all PNP germaniums and can be noise sources. Old 'lytics in the signal path have potential to upset bias voltages and cause circuit havoc. As you surmised, changing to silicon PNPs will require some bias resistor changes. My A-4010S does exhibit a tiny bit of hiss (but is quite un-noticeable when playing a tape), so I built up a part of the playback amp circuit on my breadboard using PNP Si transistors, and got it to work well, so it *can* be done in theory. It's just a matter of finding the proper biasing resistors (and choosing nice quiet transistors). The goal, in my mind, would be to get the circuit to work with just part changes, not a change in circuitry (requiring alteration of the PC board). Staying with PNP is a better plan as well.

Was going to ask if you heard the hiss in Source, but if you monitored direct from the playback amp on the deck chassis and heard it, you know where it's coming from. Got a scope? Nice thing is, since that card can be unplugged and come right out, working it over will be a cinch.

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

Will that will go into my pile of things to do at some point. That pile is getting pretty big lately!

I had a thought, maybe a "bad thought". Since that card does just snap right into the slot, maybe an all out replacement card with new caps, resistors (they drift as well), low noise transistors, and well whatever else there is... Not a plan. Just a thought.

 

[rojoknox]

Greetings from RojoLand!

Blanket-replacing the original resistors with 1% metal-film types could go a long way toward quieting the hiss, particularly in the earliest stages (nearest the head). No sense in amplifying resistor noise! And that can be done on the original board. A completely new replacement playback amplifier would be a wonderful engineering project (not just the circuit, but the board as well, considering space constraints). Now, a board could be designed and built with SMT components and perhaps be half the size or so, but there's still a relay on that board (smaller relays exist, to be sure, and there are other switching options). Power supply is positive-ground, and changing that would involve other circuit boards in the set so I wouldn't advise changing it. But yeah, a fun thought indeed.

When I rebuilt the electronics in my Concertone S510-R (a tubed machine), I swapped the fugly Suzukiohm dogbone resistors (!) out for metal-film parts.
BI-I-I-IG difference! (Set still has original tubes in great condition, including some Telefunkens!)

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

Greetings from RojoLand!

Recapping the playback amp card is worthwhile. The transistors are all PNP germaniums and can be noise sources. Old 'lytics in the signal path have potential to upset bias voltages and cause circuit havoc. As you surmised, changing to silicon PNPs will require some bias resistor changes. My A-4010S does exhibit a tiny bit of hiss (but is quite un-noticeable when playing a tape), so I built up a part of the playback amp circuit on my breadboard using PNP Si transistors, and got it to work well, so it *can* be done in theory. It's just a matter of finding the proper biasing resistors (and choosing nice quiet transistors). The goal, in my mind, would be to get the circuit to work with just part changes, not a change in circuitry (requiring alteration of the PC board). Staying with PNP is a better plan as well.

Hi J. E. (and everybody else),

I've been listening to tapes, and had forgotten just how bad the hiss/windy noise was from my A-4010S.

So I'm thinking more about replacing the Ge transistors with Si. The SM has a very convenient "DC version" of the schematic:


It looks like (apart from hair-splitting) all of the quiescent voltages are independent of Vbe, so there really shouldn't be any changes required to the resistor values to switch from Ge to Si. Do you agree? That would make this a much easier project!

The next question is which Si transistors to use. Do you have any suggestions? As well as my eyesight can tell, the original ones are:

Q1, Q2: 2SB486 (Pc = 0.12 W, hfe > 200, ft 3 MHz, Ic max 0.05 A)
Q3: 2SA49 (Pc = 0.06 W, hfe > 30, ft 5 MHz, Ic max 0.005 A)
Q4: 2SB189 (Pc = 0.25 W, hfe > 70, ft 0.5 MHz, Ic max 0.25 A)
(and the capacitance multiplier)
Q9: 2SB415 (Pc = 0.20 W, hfe > 70, ft 0.4 MHz, Ic max 1 A)

I don't have much experience in choosing replacements, but based on some web surfing and convenience (what I have on hand...) I was thinking of:

Q1, Q2, Q3 -> KSA992 (Pd = 0.5 W, hfe > 150, ft 100 MHz, Ic max .05 A)
Q4 -> 2N3906 (Pd= 0.625 W, hfe > 100, ft 250 MHz, Ic max 0.2 A)
Q9 -> ZTX790A (Pd = 1W, hfe > 300, ft 100 MHz, Ic max 2A, Vce 40V)

I wasn't able to find any info at all about the noise performance of the Ge parts. Comparing noise specs is tricky, but I just assume that the KSA992 (an "audio-frequency low noise amplifier") is a good choice. And since Q4 is the final gain stage I assume that it's noise performance isn't as critical. The ZTX part for Q9 I just picked using Digikey filters.

The two issues I see with these choices:
1) Ic max of the 2N3906 is 200 mA. But in this application (2.2k collector load) it will never get close to that.
2) ft of all the Si transistors is much higher. I've heard of people adding caps from C to E to prevent oscillation, but I't rather not modify the circuit board. (Maybe add them "dead bug" style if needed?)

rojoknox, do you remember what Si transistors you used on your protoboard experiment? Did you have any oscillation issues?


I'm not committing to a plan, but right now I'm thinking of...

1) replace the 10 uF coupling caps, listen to sound.
2) replace the other ecaps, listen.
3) replace Q1, listen. And look at output on scope for oscillation. And keep the original part.
4) ditto Q2
5) ditto Q3.
6) ditto Q4.

I didn't mention anything about resistors b/c I don't have a supply of metal film resistors. If all goes well, I'll make up an order and swap those out as well.

Ideas, comments, face palms ?

FWIW, here's the complete schematic.

 

[rojoknox]

Greetings from RojoLand!

Resistor changes are required when changing from Ge to Si transistors. The reason is that a germanium PN junction has a lower forward voltage drop (about 0.3 V) than a silicon PN junction (about 0.7 V). I will need to look at my breadboard to give you more specifics. The transistors I used were unmarked generic junkbox parts that I tested for best Hfe (I think they may have been Radio Shack 2N3906 equivalents; hard to tell). Will get back to you on that. A more appropriate part would doubtless perform better. I also don't recall any oscillation problems — was looking at a scope trace so if it were "squegging" I would have seen it.

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

Greetings from RojoLand!

Resistor changes are required when changing from Ge to Si transistors. The reason is that a germanium PN junction has a lower forward voltage drop (about 0.3 V) than a silicon PN junction (about 0.7 V). I will need to look at my breadboard to give you more specifics. The transistors I used were unmarked generic junkbox parts that I tested for best Hfe (I think they may have been Radio Shack 2N3906 equivalents; hard to tell)

The difference in the BE diode drop (.7ish vs .3ish) can make an extremely big difference or a very small one depending on the circuit.

In a "textbook" common emitter amp the base voltage is fixed by a voltage divider (and the power supply rails), the emitter is "one diode drop" lower, and that voltage combined with Re determine quiescent current. As an extreme, imagine if Vb were fixed at 0.35 V. Swap in a Si transistor and there will be no (ridiculously tiny) output.

But this circuit is different. The quiescent levels are set by negative feedback. If Vb1 were, say, a tad higher than it "should be", Ic1 would be high, so Vb2 would be too low, so Ie2 would also be low, which would lower Vb1 : problem solved.

I haven't written out the equations (or fired up a simulator) but I'm pretty convinced the change from 0.3ish to 0.7ish volts in Vbe has a very small effect on quiescent current.

I agree that some tweaking of resistor values is required. But I think that due to the circuit design (negative feedback) the existing values will work with Si transistors as a starting point. Not perfection, nut no smoke.

The real issue, as far as I can tell is figuring out what the quiescent current should be for the new transistors. At least there is more data (and designs to copy from) than the old ones.

 

[js22]

I wrote most of the previous post last night. When I was almost done my phone did two things: it up and died.

Anyway today (after some phone shopping) I did some recapping. Replaced all the 10 uf coupling caps and most of the others. No improvement.

 

[rojoknox]

Greetings from RojoLand!

All I can tell you is that the circuit did not function when I simply replicated it on the breadboard with the original resistor values but using Si PNP transistors. (Like you say, though — no smoke!) It was necessary to adjust certain resistors (not many though). Also, I did not replicate the *entire* circuit, only a portion of it. Choice of replacement transistors will have a bearing on the need for bias-resistor changes.

Now, the not-so-great news — I thought I had written down some notes on this thang, but they have apparently gone to an alternate universe. AAARGH! I've got the breadboard right here in front of me, and I'll need to take time to re-trace out what I've done. Sorry 'bout that. I *can* tell you the part of the circuit I re-did was the head amp (Q1, Q2 — both 2SB486s originally as you mentioned). That part of the circuit would surely be the most likely source of hiss.

The two transistors I used on my breadboard are National Semiconductor parts marked "942 540". Near as I can determine, they are basically PN2907s. I think I will try this again using a more suitable low-noise part (which I'll have to acquire), but I was able to get plenty of signal out with the general-purpose transistor.

In the meantime: If you ground the + lead of C12 (junction of C12 and R1; NOT the – lead!), does that affect the hiss? (left channel) Hiss *should* be lowered or eliminated if you ground the high end of VR1 (or + lead of C5).

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

I tried the test-grounding ideas. That is very helpful!


I grounded C12+ (and C27+), essentially grounding the input to the board. The noise was still there. From this I can conclude that the noise does not originate off the board (pickups, wiring,... ) Right?

Then I shorted VR1 and VR2, thus removing any input to the the Q3/Q4 part of the amp. This made the noise go away. Not 100% but a whole lot of it. So this identifies the problem as something in the Q1/Q2 sub-circuit. Right?

Also, I lied in my previous post. I did not in fact replace all the 10 uF caps. I replaced all the "normal" 10 uF caps. But not the weird 3-legged ones; namely C12/C27, right at the input stage. Once I explored a little, I found that the third (top) leg is connected to the case which is tied to ground, but not connected to either the + or - cap leg. I replaced these two caps with "normal" 2-legged ones. This did not change the noise. (Obviously, I removed the jumpers from the previous tests...)

I'm thinking next I might try to prototype the "DC part" of the Q1/Q2 circuit i.e., omit any branches with a series cap. See what the quiescent voltages are and from that infer the quiescent currents.

 

[js22]

Doh! I keep forgetting to press "Post Reply". Previous post was from yesterday.

Today I fired up LTSpice and tried the DC part of the Q1/Q2 circuit using KSA992 transistors. It worked. Quiescent voltages very close to the ones in the SM.



I never used LTSpice before (peeked at it once or twice). It's pretty cool.

Hard to read from the pic, so here's the gist:

SM ===> LTSpice
Q1 emitter -1.1 V ===> -1.16 V
Q2 base -4.4 V ===> -6.63 V
Q2 collector -10.0 V ===> -10.66 V
Ic1 -.510 mA ===> - .539 mA
Ic2 -.893 mA ===> - 1.15 mA

 

[rojoknox]

Greetings from RojoLand!

You've definitely confined the noise source to the first two stages (the head amp). C12 and C27 are shielded 'lytics which, as you can guess, are needed for hum prevention. It's easy to replicate that with a regular 'lytic by adding a sheath of adhesive copper foil tape to which you solder a ground lead (before applying the tape to the cap, of course!). I've done that with similar (but not 'lytic) input caps on the head amps of old Webcor 210s. They used shielded paper caps on the grids of the first tube (6SC7) and the thing hums without a shielded cap.

You have the advantage on me with LTSpice; I've never been able to get those simulators to work right. (I use Macintosh computers so that is also a complication.) But a KSA992 is what I was considering as well (the complement to NPN KSC1845).

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

It's easy to replicate that with a regular 'lytic by adding a sheath of adhesive copper foil tape to which you solder a ground lead (before applying the tape to the cap, of course!).

Cool idea!

You have the advantage on me with LTSpice; I've never been able to get those simulators to work right. (I use Macintosh computers so that is also a complication.)

Yeah, I run Linux so I'm used to things not working. Oddly, though, LTSpice runs great under Wine: a simulator running in an emulator.

a KSA992 is what I was considering as well (the complement to NPN KSC1845)

I've googled around and can't seem to find any compelling reasons to use anything else. Plus I have some on hand.

 

[rojoknox]

Greetings from RojoLand!

The thought occurs — consider also changing out the resistors in those head-amp stages in favor of 1% metal-film parts. That should also improve the noise issue.

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

I got in there today and replaced a bunch of resistors. I'm working on the R channel only so I can compare R vs L. I've replaced Q1-3, most of the elecrolytic caps and most of the resistors w/ metal films.

The noise is still there, but I'm beginning to notice that the noise in the right channel has a clarity, a presence, and level of detail that I never noticed before; a punchiness to the bass and a diaphonous légèreté in the upper range that I've only ever experienced in dreams. It almost feels wrong to call it noise.....

So yeah, starting to think maybe I can just live with it.

 

[rojoknox]

Greetings from RojoLand!

Perhaps you're becoming a noise connoisseur? A noise-o-phile?

I meant to ask, too, if both channels were (before the part changes) equally noisy.

Légèreté ? [lightness]

Perhaps I can sympathize. I collect and enjoy 78-rpm records. Scratch-HISSSSSSSSSSSSSSSSSSSSSSSSSSSSS.

Take care,
—
J. E. Knox "The Victor Freak"

 

[js22]

Before I started any work, the R and L noise levels were the same.

I've had some struggle between wanting to be systematic vs wanting to do what is easy and immediately feasible.

I ordered replacement electrolytic caps a while ago, so I had those on hand. Or so I thought - I was missing some values and had insufficient quantities of others. So for the caps I just went through and replaced all the ones I could, without trying to determine which were R or L channel. In particular, several of the 50 uF caps are still the old ones. All of the 10 uF are new.

For the transistors, I did R channel only, replacing Q5, 6, and 7. Then I started on the reisistors and replaced all the ones in the R channel related to Q5/6 with metal film resistors. I also replaced the one R that's shared - the one on the capacitance multiplier (R50).

Those old gray carbon comps literally fall apart with a little heat. I'm being careful b/c the foils on this board lift pretty easily, but the resistors are even more sensitive! I never opened up a carbon resistor before to see how they were made; now I know....

I started out thinking that there would be one smoking bullet to fix the problem (bad metaphor?), or at a series of clear steps of improvement. Each time I listened, though, I didn't hear any difference from before. But I am wondering if maybe there has been a gradual improvement that I just didn't sense at each step. I do recall that at the beginning there was a rumbly "wind blowing on a microphone" component to the noise. Now it's a straightforward white noise hiss. And I am cranking the volume pretty high. So maybe at each listen I've been expecting silence and not noticing tiny improvements?

Still, though, R and L are pretty much the same, so replacing the transistors and all but one of the resistors seem not to have mattered. I will definitely replace the L channel parts (for symmetry!) and order more caps to finish the job. But at this point my dreams of silence are being dialed back a little.

 

[rojoknox]

Greetings from RojoLand!

As much gain as is needed in a head amp, to achieve silence would be a near-impossible goal. But noticeable hiss at a normal listening level is not good and should be fixed. When I rebuilt the tubed amps in my Concertone S510-R, with metal-film resistors and film caps, it made a significant difference. (The original tubes, some of which are Telefunkens, tested great.) Replacing Q1/5 and Q2/6 in the A-4010S with low-noise parts should have made the biggest difference, with further improvement from changing out the crappy original resistors. But your experience seems to say different. I'm not gonna change one thing in my deck until I've breadboarded the circuit and confirmed an improvement. (Also need to cobble up a 27-volt DC supply.) Hmmmm.

Take care,
—
J. E. Knox "The Victor Freak"