GroupDIY Bridge compressor.

[Kingston]

there is a measured 346V at the input to that regulator.
The toroids are rated at 50VA each so they don't work up a sweat in this amp.  If the transformer has a regulation of 10% (which is good) you get 10% more voltage than the stated output.

I suspected it must be something like that, so you actually do have about 250VAC at the secondaries, which is pretty optimal. But not all transformers/manufacturers give such ample voltage "headroom" when used way below their recommended rating. I've had my share of bad luck in this area and tend to always over-specify generously.

 

[alexc]

Great work Dave.

I'd very much like to try this one in the future.

And always good to learn about your tube choices - being new and interesting ones to the more run of the mill.

Cheers

 

[DaveP]

Hi Alec,

The little blue Edcor has arrived, so the fun begins!

To 3U or to 4U, that is the question?

I've managed to get the FR flat from 20K to 100Hz and then it declines to -1.6dB at 50Hz, is this likely to be OK?  It drops dramatically after that, but I may be able to tweak it further down.

I can work on the layout now, as I have all the main lumps,  The number of tubes is also settled at 8!!!  I could make that 7 by losing the two 6SJ7's for a 12AU7/ECC82 but I won't as they are the only parts that are pure 26C.

best
DaveP

 

[lassoharp]

I've managed to get the FR flat from 20K to 100Hz and then it declines to -1.6dB at 50Hz, is this likely to be OK?  It drops dramatically after that, but I may be able to tweak it further down.

Maybe look at making the high end roll off at a complimentary point and with a similar roll off curve for tonal balance.  I suppose it will take some listening tests to see if it's an issue.

Are the interstages still the weak link there?  Edcor response good?

 

[shabtek]

are the polarity marks accurate on the bridge transformers in the schematic on page 2?

edit: should be 'head to tail' no?

i've been messing with a transformer bridge but cannot get g.r. no matter how i connect the 4 coils--unless a short a coil--am I wrong in thinking the active reduction element is shorting the signal together?

need sleep  now

 

[DaveP]

are the polarity marks accurate on the bridge transformers in the schematic on page 2?

Shabtek,

The polarity marks are accurate and similar to the original schematic, where black and yellow are the polarity marks, it only works with two separate interstages though.

The tube does not pass signal, it functions as a variable resistor.  The tube needs the lowest resistance when fully on and the highest resistance when the grid goes negative.  The slope of the plate curves indicate this, they have to be near vertical at 0V and near horizontal at 0 mA.  The "off resistance" has to be of the same order as the 150k resistors on the windings.

The tube grid is at -0.26V on normal levels but on peak levels can go as low as -8V, this takes the tube resistance way up which "unshorts" the windings leaving them in antiphase.

I usually sort these things out after a good nights sleep!

Lassoharp,

I shall know more about the weakest link later today, I haven't wired up the Edcor yet.

best
DaveP

 

[DaveP]

Tried out the Edcor OPT today, got +51dBm, a clean 40Vrms into 680 ohms, that's 2.35 W and enough to warm up most front ends and maybe even mad enough for EMRR.

Made some progress on the Frequency Response, I changed a few values and added a res and cap, so that it's now flat to 50Hz.  A little more work tonight and it may be flat to 40Hz, I'll update Stage one with the mods when I've finished.

best
DaveP

 

[shabtek]

nice--that edcor.

thanks for the reply re: the bridge transformers, i gotta muck around with it some more.

this is one of the most refreshing threads for a long while.

edit: (after sleep) ok I think I got it figured out: when the 'crossed over' connections between the 2 interstage xfmrs are shorted, signal passes--when not shorted (high resistance) signal does not pass (I'm just messing with 2 transformers in the bridge config. no active tube or resistors involved yet)...
If this is indeed the correct behavior than (my interpretation the phasing dots on) the schematic posted on page 2 is confused--
the dots indicating the polarity of the lower 2 coils inside the bridge should be flipped.

thanks for sharing your efforts DP

 

[DaveP]

Thanks ShabteK,

The last part is attached, Stage 3.

I have nominally put 0.1uF caps feeding the diode bridge but I'm not sure if they are the right values.  If they are too big they may slow the attack time, but if too small they may not rectify the bass strongly enough.  I'd welcome some input here.

The minimum attack time is not easy to calculate, because I don't know the output resistance of the bridge for sure.  It does need that 2.7k resistor to prevent the timing cap becoming a relaxation oscillator; with that value it is totally stable.  If the output resistance is about my 2.2 k estimate, then together with the 2.7k they make about 4k which corresponds to 1mS.  This is as good as it gets.

I'll be starting on the prototype next, it all works as a birdsnest so a tidy layout should be better.

Thanks to all for the encouragement.

Edit.  Won't be the last time I forget an attachment!

best
DaveP

 

[DaveP]

I have found that a slight redesign is necessary.  It is not possible (for me anyway) to arrange the circuit, as designed, to totally avoid oscillation due to feedback.  I have shared this problem on the drawing board and only EMRR has responded so far.

The problem with feeding the CV generating circuit from the signal AFTER the GR section is that it can "pump" the timing capacitor and oscillation starts at about 4Hz.  It works like this, the signal triggers the CV which charges the cap and reduces the signal, the cap discharges and it starts all over again.  At 10kHz and above it is much harder to get this to happen, but at 100Hz it happens all the time.

Feeding the CV generating circuit from the input to the GR section cures the problem.  This "feedforward" technique is used very successfully in the LA-2A.

I'll post these revisions next week.

best
DaveP

 

[shabtek]

are you using full bandwidth program material for these tests\oscillations

 

[DaveP]

No, just sine waves, do you think that is the problem?

best
DaveP

 

[DaveP]

I've changed the capacitors and damping on the Bridge transformers and managed to extend the frequency response to the original spec.  It is now flat from 30Hz to 50kHz but it has a slight resonance at 40kHz which I am continuing to work on.  Still whatever I do it won't affect the bottom end.

I've got the CV stage connected back to the bridge output and it now does not pump.  I'll post the revised schematic tomorrow, I'm beat now.

The timing settings affect the maximum compression possible, e.g min attack/max release allows 16.1dB,  other settings result in 14dB, the original 26C was 13dB max.

Question; The 26C has a compression in/out switch, do people think it's still worth fitting?
It does give an instant idea of the effect your making I guess, lets have some input on this before I go to metal.

best
DaveP

 

[tommypiper]

Question; The 26C has a compression in/out switch, do people think it's still worth fitting?
It does give an instant idea of the effect your making I guess, lets have some input on this before I go to metal.

I'm guessing you mean something other than a bypass switch, meaning still hearing the circuit just no compression?  That would be very, very useful.  Bypass switches can be put in easily, but hearing compression in and out within the circuit is very useful.

 

[emrr]

The compression in/out switch in an original 26C sounds vastly different.  I think they put it where they did for audio quality reasons: it bypasses two interstage transformers.  Today I'd probably keep the audio path the same for both states, and disconnect the SC. 

 

[DaveP]

Late night Doug??

I have attached Revision 1.

As the 26C switch is just a by-pass I'll leave it out, forum members will be capable of putting one in if they wish.  I'll try and fit a compression in/out switch.

I changed the 0.25uF timing cap to 1uF (along with resistors) and this cured the pumping/oscillation.

I might need a resonant trap to cure the 40kHz peak.

best
DaveP

 

[DaveP]

I've managed to lose the 40kHz resonance with the attached trap.

Later today I'll post the revised frequency response with this fitted, it's now flat  30~30kHz.

This has also got rid of most of the square-wave ringing which had me worried.

I had a bunch of coils left over from the Pultec I made....knew one would come in handy one day.

best
DaveP

 

[DaveP]

Here's the new frequency response with 40kHz trap fitted.

Blue without trap

Red with.

best
DaveP

 

[plumsolly]

Hey Dave - Excellent work! I think lots of us are watching in the wings and are quite impressed! What did you use to do those frequency response measurements?

Thanks,

Ben

 

[DaveP]

Hi Plumsolly,

Thanks for the thumbs up!

I have quite simple gear, just a sig gen and a scope, I took the measurements manually and put them into Excel, then printed as a pdf.  Sometimes Excel adds its own little curves, like at 30Hz, which aren't really there, but its just trying to smooth things over.

Simple though it is, I wish I had the gear I've got now back in 1970, when at 20 years old, I tried to make a 4 head tape deck, it all failed horribly because I couldn't see what the bias was doing, or was even the right frequency!

It might have worked otherwise, I had 2 stereo heads on a custom mount with half inch tape which cost a fortune.  The mechanics worked but my electronics sucked!

Everthing today is so much easier than back then, the internet has transformed everything, got a problem...wiki has the answer, want a component spec....its online, want parts...here next day!  It would have all taken hours on the phone to people who thought you were a nuisance, endless letters, hours in the library reading books that never told you what you wanted to know.

As far as info goes, its a different world.

best
DaveP