670 signal amp/control amp woes

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

skal1

Well-known member
Joined
Jun 4, 2008
Messages
1,282
Location
Birmingham,uk
Hi

i  have just finished building the 670 signal amp with  4 valves a side , and i was wondering what class it should be operating in  class A or AB1


input TX  600:10k

output TX :10k:600

at the moment i am using the 670 biasing  configuration -6.2Rms  on the input tx centre tap to the grid's and -6.2v Rms and ground to the dual 100 pot with the 680 cathode resistors.

The tube i am trying to use  has an Anode voltage of 250v Current of 8mA and  Cathode resistor of 250 ohm s ,this is 1 tube  so my thinking is that i will have to divide the current my 8 and i will have to scale the cathode resistor by 8 as well , am i missing something .

cheers

skal





 
ok i have to leave the signal for now , and move onto the control amp.

I have been reading Dave p and alexc PMsix61 Limiter DIY posts and it seems to be a little bit easier than the signal amp to build, so i am  going to start here.

cheers

skal
 
update


i have put together the  Class B 12ax7 and CLass A 12bh7  amps of the side chain. input tx setup 1:4  with 100k across each secondary with 24k in parallel with the  each half of the secondary.

The 12ax7 seems to be working as aspected , the only issue i have is with the  CLass A 12bh7 driver for the final , i'am a bit down onB+ on that schematic  calls for  440v but i only got 390v  so i will have to change something to get the voltages in line , don't know what  ,but i think i may have to change the plate and cathode resistors to get to 2.4v at the cathode.

cheers

skal
 
ok ,

still on the control amp right now , i  would like a little  help at this point from those, who have gone before.

so this is were i am at ,the 12ax7 class B pre is working fine and so is  the 12bh7 driver stage the problem is with the finals.

The input voltage to the  6gk6 pentodes  is  just over 60 vac  grid bias -24v , output transformer  4:1  10K:600  with a 3k9 resistor cross  the secondary and the bridge .

the signal  at  the output of the bridge looks to me a bit  mangle , i am  not using any feedback ,at this moment.

SEE Attachment of mangle wave from finals across secondary winding

http://i60.tinypic.com/18zouf.jpg

just guessing at this time , it could be a number of thing  any help from the forum would be helpful

cheers

skal
 
ok ,  so i have come to the the conclusion , that i need a bigger side chain output tx  i think i going with the one Dave p  used.

I will update when i receive the new tx

cheers

skal
 
Make sure you use a diode bridge with diodes that can take some abuse!

Little ones will blow pretty easy, being connected to a goodly powerful control amp. That will make your bridge signal wierd.

The other thing is that the control amp is biased in an interesting fashion, with the pot - that takes a while to get familiar with.

And yes, your control amp output traffo should be able to handle some 10W minimum.
ie. 10K:600 or so, 15W ideally. 20-30W is also OK but overkill.

The hard part is the isolated tertiary winding for feedback
ie. the Hammond as described by DaveP and used by me too.

wrt to the bias of the signal amp, you bias the cathodes to the voltage that makes the tubes on each phase pass a dc current  which is at the upper end of safe power consumption for your tubes.  That's your 'idle current'. You want it high.

A beefy traffo here helps - should certainly handle around 18mA per phase.  (mine is 45mA !)

Then you add the CV voltage to the grids, as required to drop those currents down to 1/30th or 1/50th at MAX limiting!
The difference in the transconductance of the parallel'd tubes in each phase of the signal amp thus determines your GR range.

Takes a while to figure out - forget conventional 'class A' or 'class AB' or whatever. 

This is bias modulation between 'max tube+traffo safe' and 'min HiFi' currents thru the output traffo.

Everything is inter-related and the end goal of a good output level  and hifi  sound  under max GR is very much component dependant and takes some tuning by ear.

I wound up with 600:15K input, running a very hot signal amp with a 10W output traffo of a low-ish ratio  8K:600 at idle 42mA per phase. CV to -82V for cutoff of GR amp with around 250ohms Zout for the control amp.

The weak point in my build is the utc a15 signal amp input traffo. It's a tiny bit bass shy and can saturate a little when slammed with a monster CV. It's also very susceptible to noise, so GOOD shielding is a must.

I ended up getting it to work pretty hood though.

As always, I'd prefer it with Sowters!
 
Thanks for those suggestion Alexc , i will take on board ,as i travel on this  journey of learning .

cheers

skal

 
Update

got the hammond tx and hook it up after some alterations to the secondary , seems to work but i am still get

ting a wave form that is not right ,could it be grid clipping  wrong bias settings to much current not enough , 

i am stump , i am using the operation condition suggested by the data sheet but no look yet....

http://www.shinjo.info/frank/sheets/135/6/6GK6.pdf


 
Where dows the signal go weird ?  ie. is it OK at the first stage, the second ?

Is it as expected with the diode bridge disconnected ?  Do you have feedback connected in ?

You need to verify each stage in the control amp without feedback, then with feedback connected then with bridge added.

Cheers

 
ok , following your debugging  Alexc  method , i narrow the weird signal down to the  2u2f ||220k||68k  circuit , when i take that circuit out i get a nice clean wave form across the .1uf  plate coupling caps ,this is with feedback in place.

what could the issue be  impedance  mismatch?

cheers

skal
 
Alec beat me to it.

Yes, remove the bridge and check the waveform across the 3.9k resistor.  What happens to the signal after the bridge is connected doesn't matter as it's converted to DC.

I set the level of feedback to give me the 100 ohm output Z that the Fairchild must have if it is to deliver its attack speed.

Use this formula:-    Zout = Rtest (Vout - VRtest)/VRtest

If Rtest is 4.7k, then measure output voltage, this is Vout.

Now put 4.7k across output in parallel and measure again, this is VRtest.

Sample figures, say 10.21V output, and VRtest =10.

Zout= 4700x0.21/10

          =98.7 ohms

Adjust the feedback resistors until you get to about 100 ohms.

best
DaveP

 
Regarding your earlier question about the signal amp.

These are the calculated voltages.



You will see that there are 16 volts across each 680 ohm that makes 23.5 mA for 4 tubes so 5.9mA /tube.

You don't say what tubes you are using but at 230V instead of 250V they will probably be in the right voltage range as they are.

best
DaveP
 
Ok thanks for input Davep and Alexc


Using this formula:-    Zout = Rtest (Vout - VRtest)/VRtest


3900 across 600 winding i get  50.7vrms

with  4650||3900  i get 50vrms

50.7vrms -50vrms =.7vrms


Zout= 4700x0.7/50

=65.8 ohms

Does this look right, this is with 3.6vrms at junction of the  12ax7 cathode

cheers

skal1


 
Zout= 4700x0.7/50

=65.8 ohms

Does this look right, this is with 3.6vrms at junction of the  12ax7 cathode


That looks great, in fact you could even reduce the feedback a little by the looks of it to increase the level of stability.

Best
DaveP
 
DaveP said:
That looks great, in fact you could even reduce the feedback a little by the looks of it to increase the level of stability.

Best
DaveP

What do mean ?    increase the resistance

skal
 
ok , the  control amp seems , to be  working  alright  now that is it has Z of 90 ish ohms so i think i will move onto the frequency response.


so what is the correct procedure  to measure frequency response of this kind of amp, and what is the realistic bandwidth , 20hz to 20khz?

cheers

skal1
 
20-20k seems like a good basis to start.

Try it with a 1W output and then at 10W.

I take readings on the scope at 10, 20, 40 100, 200, 500, 1k, 2k,  5k, 10k, 15k, 20k, 25k 30k.

Then I plot them out in Excel using a log scale, 10 minimum.

best
DaveP
 
Back
Top