Re: 807 (maybe) amplifier build.
 

:laugh2:
The pots are for setting the phase splitter balance.
Its a cross coupled design so it definitely helps to have equal currents each cathode. The pots are 2000 ohm 3 watt wirewounds. The power supply delivers 375 volts @ 1amp ;D.
If your asking about the black and the blue "boxes" in lower right underside view they are volume and balance. They are both Alps audiophool components.
The blue volume control is known as a blue velvet,, and the balance control is a black satin $125 :o the pair. I do like them though!!! the accuarcy between tracks is almost immeasurable. the balance control has a centre detent, and the tracks only go "half way". so that from centre you reduce one channel only, rather than increase one channel and reduce the other.
If you want really "hi-falutin" controls I have switched attenuators of two types, one costs me $150, the other is $450. And guess what??? they work JUST like volume controls :blah: But some people want them, and who am I to refuse??

Joe

Re: 807 (maybe) amplifier build.
 

Andy, I hope you're able to put some of this following info to good use.

I haven't really done much exploration into the LTP P/I's so decided to invest a few evenings into uncovering any pitfalls that might present themselves during construction and testing.
I decided to go with the KISS principle and work my way up from there.
It turned out that KISS isn't too bad !
Using a low-mu triode in a LTP isn't favoured by many but to adhere to the KISS principle I had to use what I had on hand.
One Compactron 6AV11 contains exactly what's needed ( 3/2 x ECC82 ) and I have a few NOS of these.
CCS is out because of KISS and didn't quite know how to settle on currents in the P/I and pre-amp.
Easiest thing: Make cathode-resistors variable ! Select values within the range you expect you're needing. Variable cathode resistor on pre-amp was chosen at 10KOhm and for the LTP I went with 100KOhm.
After rigging the test-bed I found these values worked very well.
When hooking up the TTi signal generator, the Heathkit THD-meter and my Iso-Tech Oscilloscope it was extremely easy adjusting the 2 pots for maximized output with minimized distortion.

Over the week-end I may experiment with substituting the P/I cathode resistor with a CCS to see if outputs will balance more properly.
Other than that I'm pretty pleased.

You will notice that the output amplitude is sufficient to fully drive an 807 PP output stage in AB1 into full output.
It even has spare amplitude for running the 807's in AB2 with a cathode-follower drive.

...all done with a NOS Westinghouse dumpster-tube, the 6AV11, for which I paid $1 before someone found out it contains 3/2 x ECC82 ;D

Re: 807 (maybe) amplifier build.
 

I took some more time to go over the data and curves for the 6AV11.
I'm not quite satisfied with the choice of quiescent operation point.
The low-mu tube needs to pass more anode-current that what I ended up with.
The THD-meter and the scope aren't telling a lie but I'll see what tale the circuit will tell with higher anode-current.
I'll have to lower the anode-resistors, possibly to something like 33KOhm, to achieve this.

Any views on this ?

Re: 807 (maybe) amplifier build.
 

Thanks T for that, much appreciated. It's unlikely I'll be able to get any 6AV11 from the UK but I have a few ECC82's. I'd come to the same conclusion about using the 6SL7 and was going to revert to using a 6SN7 and had settled for 33k anode resistors. I did some work yesterday drawing load lines for a 6SN7 LTP PI and AC coupled cathode follower but it looks like I may not need a CF. I can follow your schematic but what are the two 680k resistors to earth for on the OP's?

Re drawing load lines; I started trying to get my head round doing a more complex/complete analysis yesterday by doing the AC load line and figuring out the impedances of the LTP and OP stage. This is tricky if you havn't done it before as all the books and papers I've read are badly written from a novices point of view. They give figures but don't tell you how they got them. For instance Morgan Jones in his book Valve amplifiers P122 - " an accurate AC load line is easily drawn... all we have to do is move a number a squares horizontally, say 100v and draw a line through the bias point" I paraphrase. No you don't! You have to work out the resistance of your two resistors in parallel, IE anode and following grid resistor, times by the quiescent current of the anode to get your voltage or number of squares which you count to the right of your anode voltage. I hope that's right. Anyway, sorry for the minor rant, finding how calculate parameters isn't easy for a novice.

I've looked into AB2 but can't find any reference to AB2 UL in all the paperwork I have. AFAIK it isn't an option in UL because all the data I have for the 807 and other valves in AB2 has the screens a lot lower than anode HT.

Thanks again T, I'll try that circuit today and report back, Andy.

Re: 807 (maybe) amplifier build.
 

Simply simulating the loading of the output signals by the grid-resistors in the following circuit.
For class AB1 they simply ARE the 807 grid-resistors.
For class AB2 they will also be the grid-resistors but obviously not for the 807's.
Still, the loading will be the same and it will be in parallel to the P/I's anode resistors. Remember if/when calculating load-lines for this !
I put them in the test-circuit to measure more true-to-life values/performance.

Before I start reconfiguring the P/I I'll do some frequency response checks to see if lowering the anode- and subsequently the cathode-resistors will have any influence on frequency-response.
A lower cathode resistor SHOULD degrade high-frequency performance in theory.
Another reason for switching to CCS !

UL is for vales specified for the same high G2 voltage as is applied to the anode.
...or, you could lower the anode-voltage to get within G2 spec's.
At the same time you'd reduce power output drastically.

Re: 807 (maybe) amplifier build.
 

For your load line, remember that the effective load is the anode load resistor paralleled with the grid resistor of the following stage, and paralleled also with the anode resistance of the valve itself.

If a valve is run with genuine constant current source as its only anode load, the gain is still finite. It's the slope of the lines of the anode characteristic and it acts like there is a finite anode load, limiting the max gain you can get. So for your gain calculations, the valve behaves as if it has an internal anode load which winds up in parallel with the more visible ones.

If you want a better treatment of valve characteristics than gets put in modernish valve amplifier books, have a look in Terman or Langford-Smith.

David

Re: 807 (maybe) amplifier build.
 

I did extensive testing of the circuit from posting #82.
First I replace the 100K resistors with 33K and dropped the P/I tail resistor to get currents up.
Imbalance between outputs rose and became significant.
NOT the way to go!
Keeping the 33K's I then installed a simple MJE340 CCS at 6,5mA.
Immediately outputs lined-up and became equal.
;D - I thought, but alas...
All testing had been done at 1KHz and I really didn't expect it to be much different with other frequencies.
Reducing input to 50HZ, 20Hz, 10HZ and everything looked fine.
Perfect balance, full output referenced to 1KHz.
Time to test above 1KHz and the world fell apart !
At around 2,5KHz one output starts to drop. At 5KHz it's significant and really not suitable for driving an output stage.
For some strange reason the low output amplitude stays stable (about 20% less than high output) till above 50KHz.
The Heathkit THD-analyzer is very happy, THD <0,2% actually even lower, close to 0,1%.
So, *** is going on ?
I build a better CCS with 2 transistors keeping current at 6,5mA; no change.
So, AC-impedance is apparently NOT influencing output.
Also, since one output is generated from a valve in common-cathode mode, the miller-effect could influence that particular output, but that would mean a constantly dropping amplitude as frequency goes up, and that doesn't happen.
I went back to the books and had a read. There's actually a good reason for the above as I understand.
In a LTP you want to keep the tail-resistor as high as possible to balance outputs. When I dropped the anode-resistors to 33K I also had to reduce the tail-resistor to raise valve-currents. NOT especially a good idea.
My conclusion would be that the standard LTP as shown in #82 is NOT what you want. You could probably balance outputs by making one P/I anode-resistor variable, and I've seen it done.
I, however would prefer to have a perfectly balanced P/I and there's possibly a way.
I'm re-doing schematics and will report back.

EDIT: Forgot to say I replaced the Westinghouse 6AV11 with a NOS RCA ditto; no change.

Re: 807 (maybe) amplifier build.
 

The path from the input grid to the anode on the same valve is a simple triode. The cathode load presented to this triode is that of the other (non-driven) triode whose grid is at a fixed voltage, so its cathode looks like a cathode follower-buffered signal ground. o the driven triode is set up for high gain. It's grid drive impedance is fairly high, and that drives into its miller capacitance. The gain rolls-off.

The path to the other anode goes via the cathode connection and the second triode operates as a common-grid stage. It does not suffer the Miller effect, and the roll-off frequency is much higher.

So long tailed pairs driven from an unbalanced input are not very symmetrical.

This was fixed quite elegantly by Dr Bailey, who recognised that the second triode was effectivel a folded cascode for the first triode. So he 'cascoded' the first triode by making it a pentode.

Where's been a lot on here about this technique and someone posted Bailey's article from Wireless World.

David

Re: 807 (maybe) amplifier build.
 

Exactly.
I have all these articles and follow-ups.
I could post them here if anyone's interested.

For some strange reason a lot of amplifiers were done with the basic LTP w/o any regards for the symmetry problem.
Why, I wonder ?
Actually I have never noticed anyone doing extensive tests on the basic LTP with modern testequipment and posting results on the web.
My dabbling has convinced me to stay far away from this construction.

Bailey is waiting and that means selecting another valve.
I have another Compactron up the sleeve to stay with KISS, the RCA 6M11.
It contains a 12AT7 (ECC81) plus a 6EW6

I'm not trying to hi-jack this thread.
The idea is still to help Andy selecting a suitable P/I for the 807 project.
If moderators disagree just split the thread.

Re: 807 (maybe) amplifier build.
 

A tail of two Arthurs.

With apologies to Dickens.

David

Re: 807 (maybe) amplifier build.
 

I can see you've been busy T, havn't had a chance to do any work with your design as things a bit hectic here. With the LTP I have, frequency response is very good from 10hz up to 20kz but rolls off about 16khz. Balance is pretty good too, my only problem is one of power loss/driving the OP stage for which I was going to use a cathode follower.
As regards balance of the LTP, from what I've read a PP OP stage is pretty tolerant of a
mismatch. I've read the Arthur Bailey article but havn't done any experimentation.

I've looked at this problem a bit and have seen various setups from P Miller's universal driver which uses a tfmr as the PI/input stage, to various designs as used in KT88 amps in the GEC booklet, as well others. Most use a CF.

Thanks again all, Andy.

Re: 807 (maybe) amplifier build.
 

Had to take a break from this amp due to domestic pressures, but have now started back on it by knocking up a cathode follower yesterday, which resulted in a little less output power, but a cleaner sinewave. I could finaggle the driver stage but after a lot of thought I'm not happy with the OP stage full stop.

The problem lies with the OPT and it's low 2K primary.Connected as UL I can't see a way of boosting OP power without using KT88's or similar, but I havn't the money for 12 and to be honest I don't like the idea of an amp with that much power consumption, it's really not green. I've looked at lot's of OP valves including say the idea of using 12 cheap less commonly used valves but in the end I've opted to put this OPT aside and use a couple of more sensible ones I have in my stash. I'll use the 1650TA for a guitar amp or I'll re-wind it to a more sensible spec.

I'll dig the OPT's out today and then establish the turning ratio. My work so far hasnt been in vane as I've learnt a great deal, especially regarding the DC conditions of an audio amplifier. After a reading marathon I'm about ready to have a crack at the AC side of things.

Sit tight for the next installment of ..... Andy.

Re: 807 (maybe) amplifier build.
 

Hi Andy, just to throw another idea in, look at the use of big beefy PL519 LOP valves.
Also look at driving them on the screen grid(G2), not G1.
These valves were optimised for pulse drive on G1, so are quite non linear with G1 drive; however, if driven through a cathode follower onto G2 the perform well.
Plenty of stuff on the web about it.

Ed

Re: 807 (maybe) amplifier build.
 

Thanks Ed for the idea, it sounds intrigueing, one that I'll look into more. Unfortunately I havn't got any PL509's to play with so ( had a few but gave to the TV chaps) will stick with 807's for reasons below.


Back working on the amp, having squared away domestic matters. This time off has given me time to suss out the OP stage. I've looked at numerous options from using 6L6/KT66/88's- too expensive. Re wiring OPT primary into two windings to give lower G2 - too complicated, and other options of the weird and whacky variety.

In the end I'm going to have to dis connect the UL taps and ditch the idea of using UL configuration. I could rig up a switch etc to switch it back in for low level listening as an option, will see what this set up sounds like - see scan pic. Knocked this up using info off 807 datasheet, which should give an OP power of 26.5w per pair of 807's at 1.8%THD. Starting at g1 the 10k and 470p cap should give a grid stopper of 39k which with pass low frequencys up but reduce Johnson noise. Havn't really found a good value for the grid leak R but 470/680k should be about right for now. G2's 1.9k R will limit G2 I to 0.140mA approx, the 1k will drop 90v and leave G2 at 270v. Wasn't sure weather to use a cap to stop blocking distortion/FB.

Will see how this circuit goes and report back; any FB negative or otherwise appreciated.

Andy.

Re: 807 (maybe) amplifier build.
 

That'll lock the screen grid at close to cathode potential, and will make a 1k resistor get rather hot. I suspect a draughting error!

David

Re: 807 (maybe) amplifier build.
 

Just realised David, doh! Will stick a 0.1u cap between G2 and K. Am I right in "earthing" G2 to K instead of ground? Its an odd one this because I'm running on fixed bias but have a to all intents and purposes 10r K resistor.

Andy.

Re: 807 (maybe) amplifier build.
 

Got 55w out of it today, so heading in the right direction. 1k R's are of course wrong, should be 18k or higher. Now the powers heading in the right direction I can start to sort out the other stages too. Will try the CCS again on the LTP etc. Just need to see if I can squeeze another 30 watts or so out of the 6 807's and that will justify at least a little, using all the power, iron and valves.

TFL, Andy.

Re: 807 (maybe) amplifier build.
 

How far have you experimented with different 'speaker' load impedances on the OPT secondary when measuring output power?

You were worried about the OPT presenting a low 2kohm impedance to the six 807s - in other words its primary inductance may be on the low side. That will only matter at low frequencies though. If you make your power measurements at, say 1kHz, the OPT primary impedance will be so high as to be out of the picture. You can then play around with different loads on the secondary (probably higher than you're using at the moment) and find the load that gives you maximum power.

If the power supply is up to it, I'd expect to see something approaching 100W output. If you can't get to that power, I'd suspect some overloading in a stage prior to the 807s.

Martin

Re: 807 (maybe) amplifier build.
 

Hav'nt experimented at all Martin, I did briefly consider putting a 8 ohm load on the 4 or 16 ohm secs, but Rob (AKA Mr Anode) told me I'd loose low freq's. I don't think that's what your talking about though. By using a 10 ohm dummy say on the 8 ohm secs is what I think
it is you have in mind. How does this help in a practical amplifier? Can you elucidate please?

Having thought about it, the pri Z has to be low for multiple/parallel OP valves. It's just that 807's are ill suited as UL valves.

The PSU is running off a 1KVA iso and variac, so enough ooomph. The pre stages were running on low HT yesterday. The problem is every time I change the set up I have to recalculate and change resistors etc. I don't always get it right first time. So the input to the OP stage was low. The LTP/cathode follower combo I have at present works really well, I can turn the volume control up to nearly max before the OP clips, therefore have plenty of drive potential.

Andy.

Re: 807 (maybe) amplifier build.
 

Theres nothing wrong with 807's in UL configuration!! I think you maybe will need small screen grid stopper/s. As with almost all parallel output pentodes you WILL need grid1 stoppers, but I reckon you will not need more than 10K if that!! For the screen grids
50 (47R) ohms will be all you need. Layout must be kind of neat though!!

Joe

I forgot to add that 807,s will work with up to 400 volts on the screens!!
There is a note from Radiotron in Australia about raising the screen grid ratings after the Williamson design was published.
I even have a photograph of several hundred valves running on a test bed with 400 volts applied, without failures!!
j.b.