Amplifier Distortion, DC-Offset, and You!

[jblmar]

Hi all;
Great post. I have a question for you EW. I adjusted my Marantz 300DC using the Dummy Loads connected to the speaker outputs instead of connecting the DMM directly to the speaker output to measure DCV. The DMM was connected to the Dummy Load wire (+) and the (-) connected to the (-) speaker connector and the Dummy Load to the untested side. To measure the other side, I reversed the DMM and Dummy Load. My question is, using the Dummy Load, am I getting a different reading than with the DMM connected directly? Also, would the use of an Oscilloscope be more accurate. The service manual says to connect the 'scope-no mention of a meter or DMM.

Measurements:
LEFT CHANNEL RIGHT CHANNEL
10 minutes (-1.8 mV) 7.5 mV
1 HOUR 8.4 13.8
2 HOURS 12.3 16.2
3 HOURS 14.3 17.5

Measurements taken with the top cover replaced. More accurate reading since the heat generated can't escape as quickly.

Thanks,
Ron

 

[EchoWars]

I doubt it matters. I prefer using dummy loads during adjustment, and often measure at the load itself rather than at the speaker terminals...easier to get to, and the difference would only be due to the slight resistance of the wire, i.e., less than a millivolt.

 

[jblmar]

Thanks EW.

Ron

 

[jblmar]

Thanks EW.

Ron

 

[Jamman]

Thanks Echo for a great informative thread.
I have a Yamaha MX 1000.
Right 11.8
Left 13.2
I would like to see if I can get it lower.
Does it have a dc offset pot? Any help would be appreciated.

 

[Kevin's Rack]

My amps/receivers:

'79 Lafayette LR5555A L: -33mV R:-35mV (thought it would be lower)
'82 Technics SA 203 L:-9.95mV R:3.8mV (a steal at a yard sale: $10)
'84 Fisher CA274 L:47mV :screwy: R: -6.1 mV (Fisher's last good amplifier (has only transistors, no darlington packs), oh well, it needs new transistors anyway, prior owner did a horrible job repairing it)
'86 Kenwood KA75 L: -16.2mV R:-20.1mV (not bad for black plastic crap)
'01 KLH R3000 L:12mV R:14.7mV (the daily driver; figured "eh, why not)

Nothing was what I expected, especially with the Fisher [Sanyo].

Hope I'm not too late on this thread

 

[EchoWars]

Thanks Echo for a great informative thread.
I have a Yamaha MX 1000.
Right 11.8
Left 13.2
I would like to see if I can get it lower.
Does it have a dc offset pot? Any help would be appreciated.

I honestly don't know about that one...perhaps a one of our Yamaha junkies might chime in...but there is nothing terribly wrong with those readings.

 

[gyusher]

??????????????????

I honestly don't know about that one...perhaps a one of our Yamaha junkies might chime in...but there is nothing terribly wrong with those readings.

My question is. . .Why??? Seems to me that it is already very low. . fool around with it and your 10-20mV might end up 30-50mV or worse. . .

 

[Jamman]

Good point! Think I will leave well enough alone.

 

[ilimzn]

Time for one of those long, boring semi-technical posts that no one here reads... If you would like to read more about the benefits of a balanced differential pair, read here... if you are listening to an amp with 100mV or more of DC offset, you have no idea what the amp really is supposed to sound like. Indeed, some amps without a differential input are actually designed to have a bit of DC at the outputs, but this is triple-rare, and I don't think anyone here owns one. (in my book it's piss-poor design, but if you can sell it WTH..)

Actually, it is possible to have DC offset at the output even with completely balanced input pairs, and in fact there are ceses (rare as they may be) where this cannot be considered bad design. A (rare) example would be the case of all DC coupled amps, from input pot to output - DC offset will vary as the pot turns and the remedy for this (input bias current canceling) can be worse than the illness in audio amps.

Even so, for most amps, getting the DC offset down would put you at least in the ballpark if not bullseye. It should also be noted that all of this assumes input pair degeneration (not what it sounds like - it is a method of linearising the input using local feedback, look for resistors in the emitters, for some reason this is quite unpopular in Japan, was then, and still is), thermally coupled and possibly paired input transistors - the latter is the less of importance, the higher the transistor gain - which is why replacing an old input pair with some newer HG transistors is always a good idea - signifficant advances have been made in this regard since the 70s!

Regarding amps that do not have an input pair, having a DC offset on the output is not normally a feature of the design but intrinsic to the design if DC coupling is desired - it is much more difficult to thermally compensate DC offset in single ended inputs. Amps using this topology are quite rare but some are well known - NAD3020, 3040 and some of the later derivatives come to mind.

One other case where output offset may not equate to a properly balanced input pair are FET inputs. FET pairing is not as straightforward an affair as pairing BJTs, and offsets are possible at the point where the input pair is most linear (that being said, FET inputs are not known for being very linear, but they have other virtues).

ZN

 

[EchoWars]

You are correct on all points (not surprising, since you taught me more than you probably realize ), but I did take liberties with some technical explanations for the sake of brevity and to simply hammer home a basic concept.

Imbalance can be due to a lot of factors, not just mismatched diff pairs, but as long as we're not talking about some failure down the line from the differential pair, it is their job to compensate for the errors. I've had a couple of amps here that used an NPN diff pair, and I've got some ZTX small signal transistors with hfe of over 1200 (at low currents, of course). Once installed, even with a matched pair, offset still was close to 20mV. Oh well, gotta try.

Yep...the Japs about never used emitter degeneration...I can't remember the last time I saw a Japanese amp with a diff pair using degeneration. They also almost never use thermal coupling. Marantz used to have this cool little metal clip that held the diff pair transistors together. I think I've seen a few Japanese amps with some sort of shrinkwrap-type of material surrounding the diff pairs, but of course this can be done only when the individual transistors are in immediate physical proximity. Otherwise, the vast majority of Jap amps rely on matching and/or some means of nulling the offset, with no regard to thermal drift.

FET's seem to be sloppy when monitoring offsets, but I do like the way they sound when used as the input stage. :thmbsp:

 

[ilimzn]

Imbalance can be due to a lot of factors, not just mismatched diff pairs, but as long as we're not talking about some failure down the line from the differential pair, it is their job to compensate for the errors. I've had a couple of amps here... Once (replacement pairs were) installed, even with a matched pair, offset still was close to 20mV. Oh well, gotta try.

Usually this is not a failure in the sense that a component is bad, but just the result of cost cutting influencing amp topology. All I can say is, current mirrors are a nifty thing ;-) - and when you can't apply one, then you are either stuck with what you have, or adding an 'external' means of correcting the offset.

Yep...the Japs about never used emitter degeneration...I can't remember the last time I saw a Japanese amp with a diff pair using degeneration.

From memory, the Sansui AU 20000 is just about the only exception, once you take away the FET input units...

They also almost never use thermal coupling...
... the vast majority of Jap amps rely on matching and/or some means of nulling the offset, with no regard to thermal drift.

Actually, it's not that bad as long as the input pair are reasonably close and not in the path of hot air from the heatsinks. Mostly it comes down to good PCB design, as input pairs are sort of naturally close together

FET's seem to be sloppy when monitoring offsets, but I do like the way they sound when used as the input stage. :thmbsp:

Recently I've been experimenting with small signal MOSFETs. Very interesting but should probably go into a private email

ZN

 

[TorontoFish]

Help my Sanyo DCX-2700K

Hello all the AK-experts here! I am new to this vintage gear world, and this is my first post. You guys got a very interesting site here (has to be good, I waited for more than 3 weeks to have my account activated). :thmbsp:
Thank you Echo for this great thread, it took me 2 days to finish all of this, but it's worth.
I bought a Sanyo DCX-2700K from local Goodwill store, and I followed the instruction from Echo, and the readings are:
17.63v (Right)
16.90v (Left) :sigh:
I remember somewhere in the thread, Echo told someone to measure DC offset with the speakers hooked up, so I did that (the unit comes with 4 original Sanyo speakers, I figured it should be safe to hook them up), and the readings are:
226mv (Right)
258mv (Left)
Can anyone here explain why the readings are so way off? Is Sanyo DCX-2700K different or this unit totally sick? Your help is appreciated!

 

[EchoWars]

Sounds like this might be a cap-coupled unit. Without speakers hooked up, the '-' side of the cap has no DC return, and so will collect a charge. This would explain why the huge voltage readings you saw with the speakers disconnected goes away when you connect 'em.

What's odd is that you say, with speakers connected, you read 200+mV at the output. Leaky cap? I dunno. I'd have to see the prints on that one.

 

[ilimzn]

Sounds like this might be a cap-coupled unit... What's odd is that you say, with speakers connected, you read 200+mV at the output. Leaky cap? I dunno. I'd have to see the prints on that one.

Either that, or it's an amp that has the output cap doubling as a bootstrap cap, or it's a single ended design and the NFB is taken from after the cap. Both always have DC offset at the output, but usually also some means to trim it inside...

 

[TorontoFish]

I do have the schematic, don't know where to post it.

Echo, ilimzn. Thank you for trying to pin point the problem. I have the original schematic for the unit, I can scan it but don't know where to put it, for you guys to see it clearly, image size will have to be big, as a new member of this board, I really don't know if I should post a large image here, or should I put it somewhere else then post a link here? :scratch2:

 

[Wornears]

Yamaha CR-820 Values

What a useful thread! Got my new DVM to supplement my analog Rat Shack meter from '85. My latest eBay CR-820 measured 6.9 mv on L channel and 7.2 on R channel. Thanks for the thorough help Echo.

 

[Analogdino]

Here are the DC offset numbers from my vintage stable - a couple of nasty surprises...

Sherwood S-402 CP integrated amp: 18 mV and 15 mV
Hitachi SP-304 receiver: 62 mV and 51 mV Not happy!
Sansui R-70 receiver: 4 mV and 2.5 mV
JVC R-11 receiver: 9 mV and 6 mV
Yamaha M-4 amp: adjusts to zero (+/- a mV or so) Great amp!
One of my Technics SA-200's: 105 mV and 100 mV Bad news!
Another of my Technics SA-200's: 65 mV and 63 mV Not so good!
Yamaha RX-V1070 Home Theatre receiver (Dolby 5.1 analog):
L & R mains: 9 mV and 7 mV
Centre: 6 mV
Rear: 7 mV

So, how do I fix both the Technics and the Hitachi? I've no schematics for either.
Cheers,
Roger

 

[EchoWars]

So, how do I fix both the Technics and the Hitachi? I've no schematics for either.

Guess that depends on how bad you want it done...certainly you have to figure out the mechanism that the amp uses to compensate for offset, meaning you need a schematic (service manual).

I don't know much about the Hitachi, but the Technics isn't a very high-end unit. That does not mean that it isn't worth the trouble, but what I'm getting at here is that the majority of lower-end units skip the ability to zero out offset in order to save a buck or two. In these cases, the input differential pair is simply replaced with a matched set, and that usually gets offset down to 20mV or so (sometimes lower).

So if you want a resolution, and the units are important enough to warrant the trouble, you are going to have to procure some service info on them.

 

[CDFixer]

I am working on a G-8700 that takes about 10min to come out of protection when it is stone cold (sitting overnight). the offset voltage on one channel starts at about 3 volts then slowly drops to about 1.5 mv after about 15 min. the other channel starts at about .9 volts and is down to about 30 mv in 15 sec or so and then in about 5-6 min settles in at about 1MV, any suggestions as to what could be causing this I suspect the FET on the driver board as the problem follows the board.