DC offset variation.

Nuuk · 2003-07-28 10:27 am

I have just done quite a long search but have not turned up an answer as to why DC offset can vary so much.

I have just built another IGC and the offset on one channel is 32mV and on the other 52mV!

I have used a wire fro non-inverting input to ground and no resistor on inverting input to ground.

I have checked the circuit and everything seems perfect so could it be a variation in the chip!

BTW I have 254K for the feedback resistor and 10K for the input. Rails are 36VDC.

Any thoughts? :scratch:

ScottRHinson · 2003-07-28 1:36 pm

Nuuk said:
I have just done quite a long search but have not turned up an answer as to why DC offset can vary so much.

I have just built another IGC and the offset on one channel is 32mV and on the other 52mV!

I have used a wire fro non-inverting input to ground and no resistor on inverting input to ground.

I have checked the circuit and everything seems perfect so could it be a variation in the chip!

BTW I have 254K for the feedback resistor and 10K for the input. Rails are 36VDC.

Any thoughts?

I just took a look at the data sheet of the LM3886 chip (I don't know what chip you are using, so I just picked this one as an example.) shows that the input offset voltage is nominally 1mV and can be as much as 10mV. If the first chip you measured had an input offset of 1mV that would lead to ~12-20mV offset due to the gain of the circuit. If the second one ws 2mV, that would lead to 24-40mV offset due to the gain of the circuit. 10mV input offset would lead to 120-200mV of offset.

In actuality this amount of offset is more than acceptable. 200mV offset leads to 55mW of dissapation in your speakers.

Scott

Nuuk · 2003-07-28 1:56 pm

In actuality this amount of offset is more than acceptable. 200mV offset leads to 55mW of dissapation in your speakers.

Thanks Scott, I am not too worried about 52mV but was curious as to what caused the difference between the two channels when they are seemingly identical.

I can always add a 254K from non-inverting to ground if I want to bring both offesets closer to zero but I'll have a listen first!

ScottRHinson · 2003-07-28 2:18 pm

Nuuk said:
Thanks Scott, I am not too worried about 52mV but was curious as to what caused the difference between the two channels when they are seemingly identical.

I can always add a 254K from non-inverting to ground if I want to bring both offesets closer to zero but I'll have a listen first!

Well I just ran some more numbers. Let's say you put a 52mV RMS signal into an average 89dB senstive speaker at 1kHz. At 1M you'd have a signal level of about 52dB. Not loud, but audible in a quiet room.

At DC you would have a physical woofer offset of approximately .075mm. At 200mV DC offset you'd have a woofer offset of approximately .25mm. Now that's pretty interesting. Let's say that woofer had a linear excursion of 3mm one way. You'd be using 8% of your woofer excursion for DC offset.

Hmmm....I'll have to think more about that one. Maybe I'll work up the reduction in dB of headroom that you'd see because of this...where did I put my excursion to output volume spreadsheet...

Scott

ronc · 2003-07-28 2:19 pm

Thats what i did Nuuk. I tried a 500k multi turn pot on the non inverted and it driffs slightly( up to .005v). So i just buy reisitors in 100 quan lots and sort thru them and find matching resistors.Even then the offset varies about .002 v.Close enough for me.There is very little difference in the sound.
ron

Nuuk · 2003-07-28 6:08 pm

Well, here are some observations!

Despite the offset, there is absolutely no sound from the speakers when powering up, and no sign of any cone movement.

Without anything playing, there is dead silence from the speakers even with my ear a few inches away. This is in comparison to my GC monoblocks that have zero DC offset (with the equivalent of the feedback resistor from non-inverting to ground) but a slight thump when powering up and a very slight hum when listening a near the speakers.

Sound quality is fine. I don't know if I was immediately impressed as much as with my first GCs but of course I didn't know what to expect then! Also, the new GC is using new input caps where the previous ones had used caps.

The sound had improved slightly over the last couple of hours and no doubt will do so more as everything burns in.

BTW - I am using the LM3875s. The circuit is the same one shown on my DD Gainclone page except that I have 75K from input to ground, 6uF blocking caps and the feedback value is 254K..

JOE DIRT® · 2003-07-28 6:49 pm

thats basically what setup I am using NUUK except I am only using a 1.5uf mkp cap because thats what I had on hand

DIRT®

ScottRHinson · 2003-07-28 7:00 pm

Nuuk said:
Well, here are some observations!

Despite the offset, there is absolutely no sound from the speakers when powering up, and no sign of any cone movement.

Without anything playing, there is dead silence from the speakers even with my ear a few inches away. This is in comparison to my GC monoblocks that have zero DC offset (with the equivalent of the feedback resistor from non-inverting to ground) but a slight thump when powering up and a very slight hum when listening a near the speakers.

Once the amp is powered up, try disconnecting the speaker and reconnecting it real fast a couple of times. If there is even a small DC offset you'll hear it. Again, the DC offset you mentioned shouldn't be an issue, as you've discovered.

Scott

ronc · 2003-07-28 7:04 pm

Do yall think you get better sound with the equal feedback resistor (to pin 7) or without it? I do notice a slight thump when i power on and a slight humm.
ron

Nuuk · 2003-07-28 9:44 pm

Do yall think you get better sound with the equal feedback resistor (to pin 7) or without it? I do notice a slight thump when i power on and a slight humm.

If I am totally honest, I don't think that I can hear a difference. I can hear a difference when using the 18K/22K combination but between the wire or resistor on non-inverting to ground, it is too close too call.

So, you either go for zero offset but with a thump and hum, or
some DC offset without them! :yinyang:

analog_sa · 2003-07-28 9:57 pm

Do yall think you get better sound with the equal feedback resistor (to pin 7) or without it?

That's exactly what i thought. My thinking was based on a design i did in the eighties: split passive riaa with opamps (5534); all stages inverting and dc coupled. To minimise the offset and for best cmrr i had impedance matching resistors on the positive inputs. With the resistors the sound gained better depth and less subjective (and objective ) distortion.
Strangely i do not like the extra resistors with the GC. The sound is softer and slightly lacking dynamics with the resistors in place, i wouldn't call it an improvement.

cheers
peter

ronc · 2003-07-29 3:35 pm

Well yesterday i pulled the 300k(matched to the feedback resistor) resistor from both channels of my test amp(pin 7 to ground) and straight wired them.Offset went to .068v first channel and .054 second channel and there is a definate hum coming from the speakers at max volume that wasent there before.The resistors are going back to their original positions.The sound quality didnt appear to gain or lose in the test.
ron

Nuuk · 2003-07-29 5:43 pm

Offset went to .068v first channel and .054 second channel and there is a definate hum coming from the speakers at max volume that wasent there before.

Well Ron, if that doesn't illustrate that there is no 'one size fits all' answer in DIY hi-fi, nothing does!

The offsets compare with mine (at least the 52mV) but where is your hum coming from?

ronc · 2003-07-29 9:17 pm

Wish i knew where the hum is coming from? I am going put the 300k back from pin 7 to ground and see if it goes away.
ron

Ricren · 2003-07-30 12:47 am

Tump!

Ron,
Did you notice a difference in the on/off "thump" sound with or wiithout the resistor?

Ric

ronc · 2003-07-30 3:22 am

Ric
Thump is still there but not as pronounced, not nearly as loud.BTW i am using battery power if that makes any diff.
ron

Ricren · 2003-07-30 4:13 am

thump 2-tweeters!

Hi Ron.

I apology for insist but I'd like to get it right:
In your case, thump is less with the resistor or more pronounced?

Reason that I ask is that I'll biamp a set of speakers and I 'd like to go r without a capacitor between the GC and the tweeter but that need a "thump free" amp.
Ric

ronc · 2003-07-30 5:03 am

Thump is less without the resistor, but developed a hum at high volume that i didnt have before,i may go back to a 4.7uf cap on the input also.
ron

Joe Rasmussen · 2003-07-31 10:39 am

Nuuk said:
I have just done quite a long search but have not turned up an answer as to why DC offset can vary so much.

I have just built another IGC and the offset on one channel is 32mV and on the other 52mV!

I have used a wire fro non-inverting input to ground and no resistor on inverting input to ground.

I have checked the circuit and everything seems perfect so could it be a variation in the chip!

BTW I have 254K for the feedback resistor and 10K for the input. Rails are 36VDC.

Any thoughts?

Hi Nuuk

If I may comment from experience, when pin 7 & 8 sees the same DC path, I've found that most of the 3875s will have less than 10mV DC Off-set. But some, only fewish, will come in higher than that, up to 20mV.

If you go with 254K, then any DC Off-set you will get will be the original value, in my case experiences 0-20mV, plus that caused by the 254K. Notice that in your above stated case just happens to be... 52mV - 32mV = 20mV.... voila! So this would be within my expectation.

The difference is basically caused by the balance (or lack thereof) in the input bipolar devices, configured as emitter followers. If these are a perfect match, then provided DC paths are the same for pins 7 & 8, the DC-Offset should be zero. But one should expect a little drift.

So 0mV is dependant on 1) same DC paths and 2) identical input currents indicating near perfect matching.

Using 254K as an example, if you get 0.032V then input current:

0.032/254000 = 0.000000125A

So if the + input has the same (matched) input current, then a 254K will give us near 0mV DC Off-set.

But in case where we have 0.052V then input current:

0. 52/254000 = .0000002047A

This is near max in the specs - 0.2uA . The difference causes the additional 20mV.

BTW, because of the isolating factor of the input cap, this does not get amplified by the gain of the amplifier, in your case 25.4 or 28dB . Otherwise you'd end up with massive DC.

Joe R.

Nuuk · 2003-07-31 2:01 pm

Thanks for the explanation Joe.

Are you able to suggest why there is no switch-on thump and no hum from the speakers with the higher DC offset?

DC offset variation.

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member

Member