Fairchild 660 sidechain amp question

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

xazrules

Well-known member
Joined
Aug 26, 2013
Messages
172
Location
Italy
Weee folks, i'd like to know if the time constant of my fairchild clone are on specs with the original.
what i've learned here is to have the attack time of the original we need an output amp of about 100 ohm impedance. i've built mine with an hammond 1650g, 6v6gt, 12bh7 ecc.
the hammond has 500, 250, 16, 8, 3.5 ohm taps from the same coil (not a tertiary winding)
how do i know if my amp has the right output impedance to charge that cap in uS range?

there are some calculations that i can do?

i can easily have -70v on the 6ba6 grids  and no sign of thump (i think) but still don't know exactly my time costant, what i know is that if i reduce the 68k (fastest release) it can go faster again and again.

thank you in advance!
 
The primary is a nominal 6600 ohms, but actually it meets 20k ohms (2x 6V6 in series AC-wise). So the impedance of the 250r secondary results in about 750r.
If you use the 16 ohms tap, it will be actually about 40 ohms, but it may lack voltage.
But if you connect between the 500r and 250r taps, the resulting impedance is very close to 130r.
 
Thank you abbey! But what I’d like to know is how do you came up with this numbers. I really to know the calculations my self.

I’ve read here from the great Dave P that the neg feedback influences the out impedance, by selecting the series resistor.
 
xazrules said:
Thank you abbey! But what I’d like to know is how do you came up with this numbers. I really to know the calculations my self.

I’ve read here from the great Dave P that the neg feedback influences the out impedance, by selecting the series resistor.
the calculation is esay, as I've shown. You apply the impedance ratio (like 6600:250) to the actual drive impedance (20k instead of 6600). This is the open-loop output Z. But actually you have to add the resistive losses (you have to measure the DC resistance of the windings)
Now if the circuit is under NFB, calculation is not easy, because it requires knowing the open-loop gain and comparing with the gain under NFB (closed-loop gain). The resulting impedance is approximately Z(open-loop) multiplied by Gain (closed-loop)/ Gain(open-loop).
Evaluating the value requires some experimentation. Actually it is much easier to measure the output Z than calculating it from the suggested various measurements.
In order to measure the output Z, inject a signal at about 2kHz in order to read a signal taht is about 20dB below clipping.
Load the output with a resistor until level drops by 1dB. The output Z is 0.12x the resistor value.
Beware that most tube amplifiers can't operate without load. then you have to use an alternate, more calculation intensive method.
It consists in making a 1st measurement with the nominal resistive load and a 2nd with twice the load (i.e. half the value=putting an identical loadresitor in parallels.
Then you can calc the value of the out Z, indicated by r in the formula, Z being the nominal impedance.

It is quite tedious, so I would recommend you use a simulator, such as LTspice, to do the job.
 

Attachments

  • output Z measure.pdf
    292.7 KB · Views: 11
i can't thank you enough, you've answered all my doubts, and even how to calculate the open loop Z. at aiken website i've found this equation:

Zout = ((Ri + Rf) * Ro) / (Ri + Rf + Ro + Ri*A)

https://www.aikenamps.com/index.php/designing-for-global-negative-feedback

pointless if i've write it here, but i can't calculate the open loop gain of the amp, because this is not a linear (don't know the right term) amplifier, actually it is amplifing only the peaks. i've tried 1V @ 1khz sine wave at the 12bh7 and at the output trafo i measured about 25V that seems low to me, and even if i switch the polarity of my dmm it vary by 3 volts.

today experimenting with the NFB resistors i've found that the more the nfb the slower the attack, and if someone don't use a separate winding for them you don't need the 2 shunt resistors. now i'm settled with series 4k7 (with 7k and max GR there's motorboating) and by ear it is faster than an 1176.

how do i know how much signal is at 20db below clipping?
can i load the output with a pot? 5k?

have i to remove the 3k9 load resistor?
 
xazrules said:
actually it is amplifing only the peaks.
No. It amplifies the whole signal, but th erectifier uses only the peaks.

i've tried 1V @ 1khz sine wave at the 12bh7 and at the output trafo i measured about 25V
Where exactly? At the secondary? Which tap?

that seems low to me, and even if i switch the polarity of my dmm it vary by 3 volts.
Looks pretty decent to me.

today experimenting with the NFB resistors i've found that the more the nfb the slower the attack,
Hmmm... When you increase NFB, you reduce the gain, up to the point where ther's not enough signal there to compress. This is probably why you think it's slower. This experiment should be done with the same compression (GR).

and if someone don't use a separate winding for them you don't need the 2 shunt resistors.
Now you lost me.

how do i know how much signal is at 20db below clipping?
Don't you know how to push an amp to clipping? You seem to have pisunderstood teh method. Read again.

can i load the output with a pot? 5k?
You could, but it would be a much lower value. I said load the output with the nominal value. If you use the 250r tap, use a 250r resistor or pot? Now, the amp will deilver several watts, you need power resistors or pots ($$$)

have i to remove the 3k9 load resistor?
leave it there. The resulting error is less than 0.5%.

Now, what I don't get is how are you applying NFB, since the hammond xfmr has no tertiaries?
 
i don't know how to quote like you're doing, sorry!

so:

"It amplifies the whole signal, but th erectifier uses only the peaks"

so i need to disconnect the bridge rectifier and take the output at the 3k9 to test it as a normal amp?
or the rectifier is the 12ax7?

i've applied a voltage at the input trx of the signal amp until i've read 1V at the grids of the 12bh7. maybe i've to apply it directly at the bh7 grids?

"Looks pretty decent to me"

so 25V x 1,414 = 35,35V DC at CV right?

"Hmmm... When you increase NFB, you reduce the gain, up to the point where ther's not enough signal there to compress. This is probably why you think it's slower. This experiment should be done with the same compression (GR)"

i was waiting for this statement, i'd like to upload somewhere a compressed drum loop with high vs low NFB.

yes with higher nfb it compress less but i'm very very sure that the attack time was in the 2 or 3 ms range, with no nfb at all it is just unstable, with the """right""" amount of nfb (4k7 in this case) it is compressing at the edge of thumping (with still a little bit of room) but with an attack of, i don't know, but in uS range for sure and no overshoot at all. and i can't get my head around why.
with 4k7 nfb the quiescent operating condition of the 12bh7 are the same as the ones published here in the EMI schematic, the more the nfb the lesser voltage at the plate of the bh7. still i can't stand why...


"Now you lost me"

it could be helpful for someone that'd like to experiment with a trx with no tertiary winding but multiple tap of the same coil. in the original in the tertiary winding there are two resistor to ground plus two to 12bh7 cathode. if someone isn't using a tertiary winding but for example a 4 ohm tap of a multitap trafo there's no need to use the resistors to ground because they are seen in parallel (i guess?) with the the 3k9 to the rectifier.


"You could, but it would be a much lower value. I said load the output with the nominal value. If you use the 250r tap, use a 250r resistor or pot? Now, the amp will deilver several watts, you need power resistors or pots ($$$)"

sorry your right! my bad! at the moment i don't have any.

"Now, what I don't get is how are you applying NFB, since the hammond xfmr has no tertiaries?"

i'm applying nfb from the common and 3.5 ohm tap! is it no good?

i will never stop to thank you Abbey, i've learned a lot from you and the other heavyweight in this forum!!






 
xazrules said:
i don't know how to quote like you're doing, sorry!
It's basic html language. Copy (Ctrl-C) the slash-quote-between-square-brackets, then paste it at the beginning of the answer you want to insert, then at teh end of this phrase, paste the same but delete the slash.

"It amplifies the whole signal, but th erectifier uses only the peaks"
so i need to disconnect the bridge rectifier and take the output at the 3k9 to test it as a normal amp?
or the rectifier is the 12ax7? [/quote] You don't even need to disconnect it, but you must make sure that the meter is floating (not grounded).

i've applied a voltage at the input trx of the signal amp until i've read 1V at the grids of the 12bh7. maybe i've to apply it directly at the bh7 grids?
No. Your method is correct but you must measure the signal at the transformer output.

so 25V x 1,414 = 35,35V DC at CV right?
Correct.



i'm applying nfb from the common and 3.5 ohm tap! is it no good?
To where? And is the other side of teh secondary grounded?  :eek:
 
i'm applying the nfb from the 3.5 ohm tap to both sides of the 12bh7 cathodes so i think that it is grounded at the 1k bias resistor of the 12bh7 cathode. right?, and taking the output from the 234 ohm tap (250 ohm tap blk/yel wire and 16 ohm tap yel wire) to the bridge rectifier.



 
xazrules said:
i'm applying the nfb from the 3.5 ohm tap to both sides of the 12bh7 cathodes so i think that it is grounded at the 1k bias resistor of the 12bh7 cathode. right?, and taking the output from the 234 ohm tap (250 ohm tap blk/yel wire and 16 ohm tap yel wire) to the bridge rectifier.
Then the secondary is thrown out of balance, so neither the rectification nor the NFB are done correctly.
 
Sure it is not done correctly but it works and it is FAST.

To clarify I’m using the 3.5 ohm tap to the upper half of the 12bh7 and the common tap to the second half. With these configurations the voltages at the tubes are on spec.  When I come back home I’ll draw a schematic.
 
xazrules said:
Sure it is not done correctly but it works and it is FAST.

To clarify I’m using the 3.5 ohm tap to the upper half of the 12bh7 and the common tap to the second half. With these configurations the voltages at the tubes are on spec.  When I come back home I’ll draw a schematic.
DC voltages are not affected by NFB.
 
Hi Abbey i’ve attached the schematic of the actual amp that I’m settled on. And it is working fast and very solid sounding.
I know that nfb has nothing to do with DC but if I lower the nfb resistor to a smaller one to 330 ohm for the example the voltage at the plate of the 12bh7 drop to about 30v in quiescent status. Is it possible that’s oscillation caused by too much nfb?

I’d used a out trx with separate windings if I could find one but the only one was this and I thought it could be cool for experimenting. With the trx I’m using what am I messing around in respect to the original one?

I’ve calculated the output amp impedance and is at about 40 ohm.
I’ve used the equation at aikens.

Open loop gain 96v
Series fb resistor 4k7
Shunt resistor 1k
Source impedance 750 ohm

[(1k+4k7)*750]/[(1k+4k7+750+(1k*96)]
 

Attachments

  • 428D9314-FA3D-4967-8285-C7B58694EDF6.jpeg
    428D9314-FA3D-4967-8285-C7B58694EDF6.jpeg
    642.7 KB · Views: 29
I had a lingering question about the Fairchild , its seems as good a place as any to drop it here .

Would it be possible to emulate digitally the function of the sidechain/cv amp and save running what is essentially a high end tube power amp to deliver a rectified dc bias voltage to the grids of your audio stage ?
a VST like control panel could easily replace the thousands worth of hardware ,you'd still have real world analog I/O and sound ,just the 'time constants' are handled in the (computer)box ,

Anyway just bouncing out an idea ,any takers?


















 
Tubetec said:
I had a lingering question about the Fairchild , its seems as good a place as any to drop it here .

Would it be possible to emulate digitally the function of the sidechain/cv amp and save running what is essentially a high end tube power amp to deliver a rectified dc bias voltage to the grids of your audio stage ?
a VST like control panel could easily replace the thousands worth of hardware ,you'd still have real world analog I/O and sound ,just the 'time constants' are handled in the (computer)box ,

Anyway just bouncing out an idea ,any takers?
Actually, emulating the side-chain in solid-state is not that difficult; it's been done.
A VST would also emultate the idiosyncracies of the tube side-chain, but I don't think it's a significant factor in the final sonic result. After all, it's just a brute-force rectifier.
But you still need a HV amp capable of delivering about 35Vrms with an output Z of about 100r.
 
Tubetec said:
I had a lingering question about the Fairchild , its seems as good a place as any to drop it here .

Would it be possible to emulate digitally the function of the sidechain/cv amp and save running what is essentially a high end tube power amp to deliver a rectified dc bias voltage to the grids of your audio stage ?
a VST like control panel could easily replace the thousands worth of hardware ,you'd still have real world analog I/O and sound ,just the 'time constants' are handled in the (computer)box ,

Anyway just bouncing out an idea ,any takers?

Thousand worth of hardware? With 250 euros You can build the real one, leave out that sowters. The most expensive is still the rack case, knobs and meters.

The next time I’d like to experiment with a less powerful sidechain and tubes that don’t need so much cv and a 1000:600 trafo as sidechain output and try a standard nfb.
It could cut costs down by at least 1/4 and no headache.
 
Back
Top