EF85 Compressor from scratch

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

DaveP

Well-known member
Joined
Nov 8, 2005
Messages
3,019
Location
France
I have finished the design and testing of the CV Amp.



This uses a pair of 6CM7 TV tubes and it can output over a Watt of negative DC power into a 1k resistor in parallel with a 1uF cap.  These $3 tubes are essentially half a 6CG7 plus half a 12BH7.

They operate completely in class A with the Edcor TX and are at ~70% of maximum dissipation.  The feedback keeps the gain steady and brings the output resistance down to 100 ohms, the same as the Fairchild CV amp.

Best
DaveP

 
I hope you are over the worst now Ian and feeling better.

I have a template in MS Paint and just copy, cut and paste.

I made the component template from the original LA 2A schematic and have used it ever since.

I find it quite therapeutic, how weird is that ???

Best
DaveP
 
This is the power supply designed and tested:-



This provides a basic supply for the control amp and a regulated supply for the signal amp.

The regulation sharpens the knee of the compressor.

All the transformers are toroidal, it's much cheaper to use several than buy specials to the same spec.

Best
DaveP
 
HI Dave , EF85 is a good choice for vari-mu compressor . I put a pair of EF85 Valvo 
into my compressor PV-670
and made the tests of work in extreme conditions:

Compression: 6 - 10 dB
Attack time : 0.2 - 0.4 msec
Release    : 40 - 60 msec

Pictures from the video:

11jrm6d.jpg


330fejn.jpg


2qsur1c.jpg


2n3pec.jpg


2ztdkzb.jpg


Video is not an example of the sound,
but demonstration of the vari-mu compressor with tubes EF85 only.

Excuse me, sound a right channel only.
Video here:
https://drive.google.com/file/d/0B5AYDQSv9pE1NG0xRDI4UjN5NVk/view?usp=sharing

  Vladimir P.
 
Hi Vladimir P

Thanks for your post, I am very impressed :)

It must have taken you as long to test and make the demo as it did to make the compressor!

I hope my effort sounds and looks as good as yours.

(I didn't know the President of Russia was a member of GroupDIY) ..............just joking ;)

Best
DaveP
 
I have been working on the design for the signal amp, this will be similar to the 660 with 4 tubes/side.

As I mentioned in the drawing board post, I have 10 x Siemens and 10 x Mullard EF85's to choose from so I set up a test to get relative data for them all.

I tested each tube with a 4.8k plate resistor and a 517 ohm cathode resistor, these values were chosen because they give 150V on the plate and 8mA current when the supply is 188V.  With 4 tubes the plate resistor is 1.2k (4.8/4).

What surprised me was the consistency of the Siemens production and the lack of it with the Mullards.  I should also say that each tube was tested with exactly 6.3V DC and the Siemens also had more consistent heaters as well, the Mullards needed much more adjustment every time to bring them to 6.3V.


These are the Siemens test results and the tubes I chose for the left and right sides are shown with the combined currents and gains which both have to balance as well as possible.

These are the Mullard results


You can see that compared to the Siemens tubes, they draw less current but have higher gain.

The gain figures were without bypass caps but with a cap they are all about 2.3 times greater.  Remember, they are triode wired with a very low value of plate resistor as in PRR's original post.

I set them up with 4/side and they balanced very well at quiescent gain and current.  I next have to test them through the range with negative grid voltages.

DaveP
 
I made one of my dangerous mock-ups with croc clips  :eek: with 4 tubes /side and although the gain and current draw were fine at idle, they went out of balance with increasing negative voltage on the grids.  There were also potential instability problems due to having over 24mA/V gm!  These are TV tubes remember.

I decided to start with a more practical approach with just a matched pair as this provisional schematic.  This is almost exactly as PRR's original post.  I have put a copyright sign on all my circuits because I don't want them ripped off by kit builders, but anyone on this forum can copy them for personal use.



I am using just a 10k: 600 Edcor for the OPT at the moment, I have some other impedances on order to try.  Notice the way the OPT is connected, there are no coupling caps to charge and no current through the primary either, when in balance, which is virtually always.  This is definitely well within the 50V spec for this OPT.
I have tested the current versus the -Vg1 voltage in the following chart:-


I have not yet tested the gm against current, but the gain and gain reduction are shown here:-


This shows that 30dB of reduction is possible with -28V on the grids, pretty much as PRR predicted.  This has happened with about 8.4mA of idle current rather than 10mA as in his example, which should make the tubes last a little longer.

The gain is lower using the OPT direct coupled because there is a degree of negative feedback going on between the phases, the loss is more than just from connecting a 10k load.  I will be testing the frequency response with the other OPT's  later.

DaveP

 
I have finished a design for the signal amp and would welcome some expert comment.

The main issue turned out to be the need for a low output impedance to drive the output transformer to a decent frequency response.  I have used a direct coupled CF for each output and its flat from 20-20kHz.  If I had used multiple tubes, then the output impedance would have varied by too much under GR, this way it is constant at <100 ohms.


I used a pair of excellent TV pentodes 6JC6A triode wired.  These frame grid tubes represent the state of the art in 1966 and give a gm of over 10 at half the maximum dissipation, the mu is 25 so this is where the low output Z comes from.

I have tamed all the grids of these TV tubes with grid stoppers to counter any parasitics, the FR drops slowly after 20kHz then picks up again at 50kHz drops and rises again at 100kHz, this from a $13.95 Edcor TX. 8)

There are no intermediate caps to charge and the 15mA of current has no problem with the output caps.  As PRR said, the voltage change with the GR tubes is only 40V, so the CF's can easily cope with this amount of movement, which is, in any case,  limited by the regulated supply.

Best
DaveP
 
I like it Dave    :)  I've done similar with a pair of directly coupled cathode followers after the gain cell pair albeit using different valve choices.  Not much of value to say but, fwiw:

Won't probably sound or measure any better but you could lose a cap by trying it with just one 4.7uF between the OPT centre-tap leads?

I used squalid-state constant current sinks under the followers but... ?

Elevate the heater for the 6JC6's.

Protect the 6JC6's by limiting the grid to cathode voltage at turn on -    Zener. Neon. 

Edit:    Actually, looking at your schem., I would be tempted to try it with the cathodes directly connected to the OPT and then use a CCS  on the centre-tap...  Maybe another 6CJ6A? 

Anyway, just thoughts, looks good as is too, cheers.

 
I like it Dave    :)  I've done similar with a pair of directly coupled cathode followers after the gain cell pair albeit using different valve choices.  Not much of value to say but, fwiw:

Won't probably sound or measure any better but you could lose a cap by trying it with just one 4.7uF between the OPT centre-tap leads?

I used squalid-state constant current sinks under the followers but... ?

Elevate the heater for the 6JC6's.

Protect the 6JC6's by limiting the grid to cathode voltage at turn on -    Zener. Neon. 

Edit:    Actually, looking at your schem., I would be tempted to try it with the cathodes directly connected to the OPT and then use a CCS  on the centre-tap...  Maybe another 6CJ6A? 
I will try one cap and no caps as you say.

The heaters will be elevated on the 6JC6A's.

The grids come up slowly with the regulated supply which gives a soft start effect.

These are my provisional timings which were also used on the Group DIY comp.



Any comments welcome

Best
DaveP
 
I have been working on the best OPT to use for the signal amp.

I tested a range of Edcor OPT's of various impedance and you might be interested to see the results.
These were tested with a common input voltage except the 600:600 which had square waves at that input level, I reduced the level for that TX until it was back to a clean sine wave, although the output level was fair, I don't think I'd like to run the amp into such a low load.  Output Z of the amp was ~100 ohms.



I measured on the scope, the points where the voltage went -3dB in the top and low ends.  This rules out the 15k:600 and the 10k:10k as the top end starts to drop way too early.

The 2.4K:600 and the 10k:600 look promising.

Best
DaveP
 
This project has now gone live.

I started cutting the metal today.



Gathering all the parts together.

DaveP
 
The front panel is now finished, this is the back view showing the tabs which connect to the rest of the chassis.


I have just thrown in the meter to show what it will look like from the front.


I have to print a dB scale for the meter.

The controls will be:-
Meter select, Attack, Release and Threshold.

Unashamedly right-handed, can't help that!

DaveP
 
Finished the chassis structure today.



I still have to plan the layout and cut the tube base holes.  There is a transformer/rectifier module to fit on the chassis, this will be made from galvanised steel to reduce the magnetic field, toroidal transformers are used throughout.  I shall be using Merlin Blencowe's grounding plan for this project.



The protective layer is kept on as long as possible.

There are 9 tubes and 7 transformers to fit in.

A little bit about the design.
It works like a smaller Fairchild and should have the same speed but with more variable timings for today's music.
It makes use of tubes at the pinnacle of tube development around 1960, especially high performance TV tubes.
The signal has only one capacitor in each half cycle and is not subject to feedback.  It could probably be direct coupled throughout, but slight imbalance could put DC through the primary so I opted to leave the caps.  I have gone for a tube regulator for ruggedness in the prototype, but if you are happy with silicon then you could use that.  But I have used the very latest Schottky diodes to rectify the heater supplies, they have around 0.3V Vf depending on temperature.  They are operated without heatsinks due to the very low load loading.  One supply is elevated to 140V to help the reliability of the regulator and the cathode followers.

Best
DaveP

 
Thanks Thomas,

I have not been up to any more work on the project as I am into my third week of bronchitis and I feel lousy.

All I have been able to do is ruffle a few feathers on the Brewery ::)

I am seeing the doctor on tuesday so maybe he will give me some antibiotics as it must be past the viral stage now.

Best
DaveP
 
I managed to screw up my doctors appointment last week because I misheard the time the receptionist said (in French), I should have been there at 11.15 but I only caught the 15 and turned up when the doctors was shut at 3pm!!  Dickhead!

So I didn't get my antibiotics until a new appointment this monday and I'm only starting to feel a bit better today.  If this happens  to people in the near future when all the antibiotics are useless we will just die I guess.

Anyway, I have not been totally idle, I have been moving components around to finalise the layout.

Thanks for your concern

best
DaveP
 
Back
Top