Cant understand...racking Mccurdy EQ (((((FIXED))))

[radardoug]

I was sent a private message about the capacitors in this module so I went back and had another look at the circuit, and was shocked by the number of electro's in the signal path. So here is my answer to the private messgae.

There are 17 electrolytic capacitors in the signal path. A lot of these have been put in to make sure that offset cannot transfer from one stage to the next, but it is very conservative design. These caps will be affecting the low end.
For a start C1 and C2 can be removed and linked across. As long as the previous device has no offset, then they are un-necessary. Then you need to measure with a good voltmeter with high input impedance across the following caps. If there is no offset then you can remove them and link across.
C7,14,16,20,22,15,30,36,42,47,52,53,59,67,68.
Even if there is a little offset, say 10 mV, you can remove the caps. You can always take them out one at a time and check the circuit is still good.

 

[abbey road d enfer]

There are 17 electrolytic capacitors in the signal path. A lot of these have been put in to make sure that offset cannot transfer from one stage to the next, but it is very conservative design.

Typical of good professional design practice.

These caps will be affecting the low end.

In a predictable manner, which the designers have taken into account. Except that lytics lose value with age and need to be replaced after some time.

Even if there is a little offset, say 10 mV, you can remove the caps.

Then you may experience switching clicks. The main reason they are here in the first place is because there is nothing more annoying than having clicks when you switch the EQ ON or you modify the boost/cut. Much more annoying than whatever subtle impairment caused by the lytics.

You can always take them out one at a time and check the circuit is still good.

Offsets are not fixed, they drift with time and temperature, you may measure zero at one time, positive at another, and negative later.

 

[3nity]

Abbey.
The connection i moved are at the input...output unchanged...
Thanks.

 

[3nity]

Edited: problem still there.

 

[abbey road d enfer]

What do you mean "shield on the input cant be connected"? Pin 3 is ground, as well as Pin 13. There is no reason that they can't be connected to the shield of a connecting cable. There seems to be something wrong there. Do all the modules have the same problem?

 

[3nity]

Yes both modules act the same.
could be that input at pint 1 is shorted to ground by R5??

 

[abbey road d enfer]

That would mean that someone has modded these modules to make them unbalanced...wacky! ???

 

[3nity]

but without shield they are balanced...

 

[abbey road d enfer]

You really need to check how each leg reacts to signal; injecting into each leg separately, with signal referenced to the gnd pin, the non-inverting input should receive signal with about 6 dB loss, and the inverting input almost without loss

 

[radardoug]

If the two legs differ in gain by 6 dB, then this input stage has extremely poor CMRR.  They should have the same gain, within 1% gives 40 dB CMRR, within .1% gives 60 dB CMRR.
Abbey road d enfer has taken me to task for my replies re the 17 caps. A couple of other points.
In a really professional design, the designer would take into account offset, and would not introduce 17 low quality components with limited lifespans into their circuit. There are ways to design such parts out, I built a very good eq circuit that was d.c. coupled except for the frequency determining components and it sounded very good. Yes, offset will vary with time and temperature, but a good design would not be affected by that. And as I suggested, take out one cap at a time and listen.
As to the comment about the cap values having little effect on the sound, if each of the caps affected frequencies of interest by .1 dB, then 17 in series is 1.7 dB of effect. Suddenly its not trivial any more!

 

[abbey road d enfer]

If the two legs differ in gain by 6 dB, then this input stage has extremely poor CMRR.  They should have the same gain, within 1% gives 40 dB CMRR, within .1% gives 60 dB CMRR.

They will have the same gain if the unused leg is shorted to ground. That's one of the caveats of this type of balanced input.

Abbey road d enfer has taken me to task for my replies re the 17 caps. A couple of other points.
In a really professional design, the designer would take into account offset, and would not introduce 17 low quality components with limited lifespans into their circuit. There are ways to design such parts out, I built a very good eq circuit that was d.c. coupled except for the frequency determining components and it sounded very good. Yes, offset will vary with time and temperature, but a good design would not be affected by that. And as I suggested, take out one cap at a time and listen.

So you think these designers at Mc Curdy were second-class? I don't and I think they had their priorities right.

As to the comment about the cap values having little effect on the sound, if each of the caps affected frequencies of interest by .1 dB, then 17 in series is 1.7 dB of effect. Suddenly its not trivial any more!

Any good designer knows that, since the beginning of wireless. If the frequency response is down 1dB at 20Hz, does it matter that it is created by one single pole at 10Hz, or by 17 poles at 1.2 Hz? And BTW, most of these CR have <1/1000th dB at 10 Hz. 17 of them make less than 2/100th dB.

 

[3nity]

Oh my gosh...cant remember when was the last time i use my scope!
Abbey it all reads ok with the scope i read 8.5db at the inverting output and 6 at the non-inverting...could this be right??
Thanks.

 

[abbey road d enfer]

It doesn't seem right; the difference should be close to 6dB. Or if you short the unused leg to ground, you should have the same level (+/- a fraction of dB).

 

[3nity]

When injecting signal into the non-inverting pin i had signal at the inverting at the same time.
but if signal goes into the inverting signal dont get to the non inverting...is this normal?
Thanks

 

[abbey road d enfer]

This is absolutely normal. When you inject on the non_inverting, the opamp acts as a voltage follower so the neg input follows the positive one, but when you inject on the inverting input, it acts as an inverter, with the positive input referenced at ground.

 

[3nity]

Thanks abbey.
I think i will return the eq to the client because i cant make them work right!
The problem is simple but not the answer.
balanced signal results in out of phase, bad quality audio and low gain.
unbalanced sounds really good.

What a shame...  :-[

 

[3nity]

Whats more weird its the way they use to work a few times..
First time i tried them they worked just ok.
Next time in rack not working take them out of rack and work ok...now..

Thanks.

 

[3nity]

Here we go again back to the beginning...
Im gonna try a line input to bypass the debalancer circuit...

 

[3nity]

So the problem isnt at the input debalancer.
I put a line input transformer and it did the same thing..

Now my theory lies in the psu section now!

 

[3nity]

Im back again.
I dying slowly here.
Still cant understand why it all works like they should when input shield is left unconnected.
that way all works ok but i have some hum that goes away if i connect shield...but then audio goes wrong...
all voltages look ok and every thing is recapped.

im gonna record some audio samples so you guys hear what i hear.
Wish me some good luck!
Thanks.