My chip amps don't travel well...

I took along two chip amps to a friend's for an amp comparison. Of course, both misbehaved!

Both went noisy - in my system AC hum measured at the speaker terminals is less than 0.3 mV on all channels of the two stereo amps. I didn't have a chance to measure at the friend's place, but my guess is that it was at least 10 times higher than this with the quietest of the two amps (his speakers are around 94db/w/m sensitivity - so 3mV is easily audible at the listening position.) The other amp had around 10 mV hum as a guess as it was substantially louder. The two amps are very different in terms of grounding - one is completely floating from the safety earth (it's on a piece of wood so it's inherently unsafe) and the other is in a heavy aluminium chassis with the circuit ground attached to the safety earth through 15 ohm NTC thermistors. The latter was the quieter of the two.

The second problem is that DC offset increased dramatically. One of my amps only measures 1-2 mV on both channels at home - at the friend's this same amp was 250 mV. The other amp measures 20-40 mV and increased to over 100 mV in my friend's system.

It is interesting that these problems persisted even with no input to my amps. So could it be that the speakers / speaker cables caused these problems? I'm not using a Zobel on the output, or a series resistor / inductor. Maybe the amps didn't like his silver foil speaker cables? I wouldn't have thought a problem with oscillation would manifest itself so obviously (or have any impact on the DC offset.) I didn't notice either amp running unusually warm - but we didn't play them for long because they didn't sound all that good.

I couldn't measure any DC offset from the output of my friend's DAC (and I shouldn't as it has a valve output stage with coupling caps.) Other amps did not exhibit these problems (an AKSA, and a Sonic Impact digital amp) in the same system.

Any thoughts?
 
Very interesting. I have measured mine many places and the DC offset was always within 1 mV everywhere. My travel problems were due to poor construction. I thought DC offset was usually an artifact of the chip, not the quality of the power. And, like you, I've got a NTC in line with the AC ground prong. KT60 or something like that.

One thing that you can do if no one here is able to help is to have your friend transport his speakers and cable to your place, and see what the measurements and hum are with that setup.

Inverting or non-inverting?
 
I suspect either RF interference (especially on the wooden one. I also like mine to have Zobels, input resistors with a small cap to remove RF, small value bypass caps at the chip pins, and sometimes I put a smallish cap accress the feedback resistor to bring the gain down at ultrasonic freqs. I also try to filter the mains if at all possible.) or construction quality. On some of my easlier homebrews, my construction wasn't up to par, and sometimes wires would rattle lose when I move them around. I doubt different speaker and power wires would make a difference

Did you just suggest making a 120 volts AC power wire out of CAT5 netowrk cable?!?!??!? sounds crazy and unsafe. I just stick with the generic computer power cords, and use the cat5 whenever I need some generic low voltage wiring that isn't going to be carrying an *** load of current.
 
analog_sa said:
Is it possible your friend's speakers are a lot more sensitive and hum becomes more audible

They are some 6db more sensitive than mine, but 0.3 mV is *not* going to be plainly audible at the listening position - which is the measurement I get at home. I didn't have the opportunity to measure at my friend's, but I know it was much more than 0.3 mV.

I know what 10 mV of hum sounds like - I've built quite a few SET valve amps.
 
About your DC-offset.

If your amplifier is un-buffed, the DC-offset will to some extent depent on what output-impedance of your preamp/DAC that the input of your amplifer it "looking" at.

Remember that if your amp is DC-coupled any "extra" DC-offset will be amplified by your gain !
(so typically with a gain of x20 and 5 mV extra you get 100 mV DC-offset !!)

Have fun

THomas
 
tlmadsen said:
If your amplifier is un-buffed, the DC-offset will to some extent depent on what output-impedance of your preamp/DAC that the input of your amplifer it "looking" at.

Remember that if your amp is DC-coupled any "extra" DC-offset will be amplified by your gain !
(so typically with a gain of x20 and 5 mV extra you get 100 mV DC-offset !!)

Yes, the first point is a big difference between the two systems. In my system I use an attenuator on the input that is only ~100 ohms at DC. The friend's system uses a pot at the output of the DAC, but I don't know the details of how it's set up.

BUT...as I said before it exhibited the above mentioned symptoms with the inputs open, too. The only thing connected was the speakers.

I know about the second point and I checked that there was no input DC entering the amp.
 
Hum and noise measuements are normally made with the inputs to the amp shorted a condition which more closely resembles the normal operation with a signal source connected (usually sources are lo-z and amps are hi-z input). Your increased hum and noise may be due to leaving the inputs open.

As for cable capacitance, putting the parallel RL device on the output of the amp will prevent any instability due to capacitive load. The one in the amp chip dtaa sheet works fine, and no, you won't "hear" the thing.

This minimalist amp stuff that seems to be the style these days has a lot of drawbacks, as you are finding out. You can leave out the "extra" parts and can build an amp that's barely stable with your specific speakers and cables attached, but then you can't connect other people's equipment to it. Those guys at National who designed these things are no slouches. They didn't pepper the circuits with "extra" parts just so that their brother-in-laws who own resistor and capacitor companies would make a few more bux. They put those parts there to ensure stable amplifier operation for nearly any load/source combinaton.

Yes, I know, I know... those "extra" parts muck up the sound...

I_F
 
I_Forgot said:
Hum and noise measuements are normally made with the inputs to the amp shorted a condition which more closely resembles the normal operation with a signal source connected (usually sources are lo-z and amps are hi-z input). Your increased hum and noise may be due to leaving the inputs open.

Yeah, I know.

This is quoted from my initial post: "It is interesting that these problems persisted even with no input to my amps." As this implies, I didn't only have the problem with no input connected. I started with an input connected, then I pulled the input to see if the hum and DC offset was upstream. It wasn't

I already know what each of the parts in the datasheet example circuit do and the dangers of omitting them. What I want to know is what problem the symptoms indicate.

The increase in DC offset is particularly puzzling.
 
The input shunt resistor of 22k limits the offset. So no matter what source impedance is, the offset shouldn't be more than your usual maximum (approx 70mV with average chip, 20mV with the better ones, at 22k input resistance).

It would be interesting to measure if increased offset also occured when ribbon cables were not connected (as analog_sa already mentioned).

I'm also using copper foil cables (8 ft runs) but they are loosely spread and not side by side. Even without Zobel, I'm not experiencing any anomalies.
 
Peter Daniel said:
The input shunt resistor of 22k limits the offset. So no matter what source impedance is, the offset shouldn't be more than your usual maximum (approx 70mV with average chip, 20mV with the better ones, at 22k input resistance).

Yeah, I get 2 mV both channels with my passive attenuator attached and 40 mV both channels without (with 22k input shunt, of course.) I measured 250mV in my friend's system on the same amp, with or without input - it didn't matter. This doesn't make sense to me.

The noise is less mysterious, I'm probably picking up some sort of EMI. In my home system, with my low impedance attenuator (100 ohms at 0 Hz) there is no problem when it's connected, but the noise increases quite a bit in one channel when it's not. At my friend's the noise was present in both channels with either his preamp or DAC connected - it didn't matter which.

Peter Daniel said:
It would be interesting to measure if increased offset also occured when ribbon cables were not connected (as analog_sa already mentioned).

I didn't have the chance.

Peter Daniel said:
I'm also using copper foil cables (8 ft runs) but they are loosely spread and not side by side. Even without Zobel, I'm not experiencing any anomalies.

I use Kimber speaker cables which are fairly capacitive with no problems at home. This guy's silver foil cables are run with +/- in the same jacket, so they would be capacitive.
 
Litz power cords

Ok maybe I should start a new thread, or admin could do it?

Wait a second - you mean using litz AC power cable is not accepted as a way to vastly improve sound quality?

Did you just suggest making a 120 volts AC power wire out of CAT5 netowrk cable?!?!??!? sounds crazy and unsafe. I just stick with the generic computer power cords, and use the cat5 whenever I need some generic low voltage wiring that isn't going to be carrying an *** load of current.

Generic computer cords sound likejavascript:smilie(':dead:')
dead So do generic extension cords.

Crazy, sure, unsafe, well, I think safer than being a pedestrian.

If anyone does nothing else this year, try this upgrade on all powered components.

The main problem with audio is most people are "visual" meaning they hear with their eyes (or alternately, what's in vogue in print).

But members of this forum might be at risk of being too logical and therefore not open to experimenting with empirical findings that may not have an obvious explanation.

I was also a nonbeliever until I tried it.

Generally you need small gages, i.e. multiple strands of 24 like CAT 5, solid better than stranded, and the more litz conductors the better (although 4/conductor is actually lots for a small chip amp current-wise, do the math)
 
jeff mai said:
I'm not using a Zobel on the output, or a series resistor / inductor. Maybe the amps didn't like his silver foil speaker cables?

You may have problems with the amp driving those cables.
Or, even discarding the cables, it may be that the amp doesn't like those speakers' (complex?) crossovers.
Join the two factors togheter and you're in for troubles.
When you make an amp exclusively thinking in your system/setup and you think it doesn't need a zobel, it may work fine, but what happens is that you can't take the amp anywhere, it just works in your system.

The amp needs a zobel, I do use it, and I also have Kimber 4TC speaker cables.
With these cables and my Epos speakers (which have a very simple crossover, just on the tweeter) I do use a zobel, with better sonic results.
I have tried simpler/cheaper cables, but something in clarity and detail is lost, so Kimber it is. With a zobel.
The minimum you can do is to put a 2.7R+100nf snubber from output to ground.
Btw, I take my amps everywhere and they perform as well and predictably as in my system.
And Btw(2) I do use a 300pf cap between inverting and non-inverting pins on the chip.
 
Re: Re: My chip amps don't travel well...

carlosfm said:
The minimum you can do is to put a 2.7R+100nf snubber from output to ground.

I thought these components were for stability, shunting away HF oscillations and not directly related to the load. The output resistor and inductor shown on the datasheet are there to decouple a capacitive load.

I put the snubber on because it was easy to do, though I am not able to see if it solves the problem because I am at home where everything is fine. My first impression is that it hardened the upper midrange and forshortened the soundstage, but who knows.
 
Re: Re: Re: My chip amps don't travel well...

jeff mai said:
I thought these components were for stability...

Yes, and that's probably the problem you experienced.

jeff mai said:
... shunting away HF oscillations and not directly related to the load.

Not directly? I wouldn't say so.
Most speakers have a rising impedance towards the high frequencies.
Some more than others.
Depending on the severity of this problem, and also the cables you use, the amp may become unstable.

jeff mai said:
I put the snubber on because it was easy to do, though I am not able to see if it solves the problem because I am at home where everything is fine. My first impression is that it hardened the upper midrange and forshortened the soundstage, but who knows.

Then try other values.
Lower the cap to 47nf, or go for higher resistor values.

Go back to your friend's house with some RC combinations on your hand and test the amp again, with these RC elements across the amp's speakers terminals.
Easy to insert, change, remove.
Also, take a DMM with you.
When you get the amp to behave on his system, it's very likely that you can travel everywhere.:D