First valve amplifier build

[Radio Wrangler]

Originally Posted by GrimJosef

Looking at the photo and the fairly primitive original design I'm afraid that the manufacturer may have spent little on the mains transformer, and it may be running quite close to its current limit. The 6K6's do seem to need significant current, albeit at relatively low voltage, to work as they were designed. So while it might have been nice to boost the HT voltage with a more up-market rectifier arrangement I was nervous that any consequential impact on the safely available current could be serious. I think I'd want to know what the mains transformer's current limit was before setting about redesigning the audio amp's output stage (other than to do away with the LTP scheme, which does need to go).

Cheers,

GJ

I suspect that going to a bridge rectifier and reducing the reservoir capacitor value appropriately would reduce the peak current over mean current ratio and reduce dissipation, increase efficiency. Rectifiers are highish stress things to start with, but half wave ones are even more so, and the transformer will likely behave better without the net DC component that a half-wave rectifier gives. It will be less into saturation, and the magnetising current-caused dissipation will be less.

David

[Radio Wrangler]

[QUOTE=Gabe001;1403849. A closer look at the transformer revealed a cut wire. It turns out that the transformer's HT secondary winding was originally centre tapped to work with a rectifier valve.

So I lashed up 2 diodes on a tag strip with a 100ohm surge limiting resistor. The transformer appears comfortable but I'm keeping a close eye on it. The lamp limiter behaved as it should
[/QUOTE]

Ah, that's a lot less scary.

Half-wave rectifiers for anything above a few mA of HT were rare for a reason.

TVs and AC/DC radios used half-wave rectifiers of course and at a fair bit of power, but then they exported the DC component problem straight back to the electricity board's substation.

Dvid

[Gabe001]

Does the capacitance value of the grid coupling cap in a push pull amp circuit change the quality of the sound?

I suppose it can affect the amplitude of the AC signal (capacitance too small=clipping), but does it affect frequency?

[G6Tanuki]

Originally Posted by Gabe001

Does the capacitance value of the grid coupling cap in a push pull amp circuit change the quality of the sound?

I suppose it can affect the amplitude of the AC signal (capacitance too small=clipping), but does it affect frequency?

Yes, like you say, if it is too small it will provide a roll-off of the low [bass] frequencies.

That's not necessarily a bad thing: indeed it was a deliberate 'trick' used in quite a few push-pull transmitter modulators used by amateur-radio types, where we concentrated on 'communications-quality' and so low frequencies [anything below about 300Hz, ahich add little to the intelligibility of speech] being cut was seen as a good thing.

[snowman_al]

It could change the sound if it was too small or too large in capacitance.
Very basically the cap and grid leak resistor act as a filter.
You want the filter to pass enough bass (so the cap needs to be a certain value) but not so much it could saturate the output transformer on very bass heavy material. Rule of thumb is the smaller the cap the less bass it can 'pass' and vice versa.

[Gabe001]

Got it. Thanks both

I found the maths here
https://ampgarage.com/forum/viewtopic.php?t=17420

So why not simply always use a 47nf or a 100nf and allow all frequencies through unimpeded all the time. I understand limiting the frequencies for speech only amps, but for music, wouldn't a wide frequency range be a good thing?

[snowman_al]

Everything is a compromise and also interrelated.
There is little point in trying to amplify outside the frequency range of the output transformer (low or high). Trying to reproduce lower frequencies than the transformer can handle just saps power and increases distortion.

[Diabolical Artificer]

So why not simply always use a 47nf or a 100nf and allow all frequencies through unimpeded all the time.

Stability. If your power supply resonates at 5hz for example you want to chose RC filter cut off frequencies to deal with that. Same goes if you HF instability with certain OPT's. It's a juggling act.

Andy.

[Gabe001]

Thanks for the replies. I should be able to post an update this weekend

[Gabe001]

Righto

Big update. I spent the weekend building the amplifier based upon the circuit below. Jerry (cathoderay57) and myself did some calculations etc

I've set up an RCA input (L+R) so I can use a stereo cable. The signal is summed via 2 resistors (180 ohm) and it goes to volume control via a 22nf cap.

Just had some time to turn it on for the first time before swmbo had enough...

Voltages are as expected.

Under-chassis view below. I probably made all the rookie mistakes under the sun. I tried to keep the AC away from signal and I use a single earthing point. I think I should have used screened cable esp as I've got some distance between v1 and V2; I may redo the yellow wire. In any case it has been great fun.

Incidentally I'm using an ecc83 as v1, but I'm only half of it. Also, I haven't set up any feedback from the speaker yet.

These are my observations so far:

Lots of hum, but if I do the following it almost completely disappears:
1. Plugging in my phone to the RCA sockets. This has the greatest affect
2. Bypassing the volume control.
3. Bypassing the 22nf cap before volume control, although this has very minimal effect

Meaning that if I direct the phone signal through the RCA sockets, the summing L+R resistors and directly into the grid of the ecc83, it goes completely silent. The hum in this case can only be heard with my ear to the speaker.

It plays really well, good rich sound which fills up the room, even with the crappiest of speakers. I'll hook up the oscilloscope as soon as my 100watt 4 and 8ohm resistors arrive and do the tests described in the bvws article.

Any thoughts, in particular why the hum is behaving the way it does?

Thanks
Gabriel

[ms660]

The twisted pair of 50Hz radiators near the input won't be helping, usually center tapped heater winding for a twisted pair or a humdinger across the heater winding as is.

Lawrence.

[cathoderay57]

Hi Gabriel, the hum pickup is less likely to be the long yellow lead to the 6C4 grid since the gain of that stage is relatively low. Most likely the pickup will be from the grid circuit of the ECC83. Try a screened lead to replace the shorter yellow wire from ECC83 grid to the volume control. Also, solder an earth wire to the volume control pot cover, unless it's aluminium. If it is try holding an earth wire against the pot cap and see if it makes a difference to the hum; if it doesn't then there's no need to bother. The only reason I can think that the phone input reduced the hum is maybe it is a low impedance source but going in series with part of a 1M pot it shouldn't be the case. Running any high gain amp stage without an input and the gain cranked up will generate hum. Try putting a 100k resistor across the input terminals and see if it still hums as badly. As Lawrence said a hum-bucker is usually a good idea when you don't have a centre-tapped heater winding. Cheers, Jerry

[Gabe001]

Sorted. I don't know whether I did the right thing, but earthing one side of the heater supply (blue wire) stopped the hum completely. So effectively I now have no hum at all

So all that is left is the noisy volume pot, which hisses between around 50-90% and then goes silent again.

Options are:
New volume pot
Resite to between v1 and V2 so that noise it generates isn't amplified much (the problem with this is that the 6c4 valve is at the back)
Remove the volume control and manage volume from phone/device only

[Herald1360]

Earthing one side of the heater supply is much better than leaving it floating!


Putting a 50-100R WW pot across it, earthing the slider and tweaking for minimum hum possibly better still- esp if you run to any more stages in front of the ECC83!

[GrimJosef]

It may not be the pot which is the source of the hissing. It may be the ECC83. When the pot is set to low volume (less than 50%) the ECC83's grid will be connected to ground by the 'bottom' section of the pot, which will be a relatively low resistance. When the pot is set to high volume (greater than 90%) the grid will be connected to the source by a relatively low resistance - the 'top' section of the pot. But the source will have a low (output) impedance to ground. So once again the grid will be connected by a low total resistance (pot top section plus source output impedance) to ground. In between these two points the resistance from grid to ground will be relatively high. When this is the case, noise generated in the valve itself may not be 'sunk' to ground very effectively. It is also true that the noise can be coming from the pot. In fact there will always be some noise due simply to the pot's resistance itself.

One solution would be to use a lower resistance pot. These days most sources, and certainly your phone, would be capable of driving, say, a 47k or 50k pot. If you can track down one of those you should find the hiss volume drops substantially. Even a 100k or 250k pot should improve things. Since it's being used as a volume control it should be log taper, of course.

Cheers,

GJ

[Gabe001]

Thanks Graeme

I'll get the order in for the pot you recommended so I can do more work over the weekend. Also, I'll try to post some tracings with the oscilloscope as soon as the speaker load resistor arrives

I'm tempted to try reducing the 220k resistor feeding the grids of the 6k6 to get more volume. Is this value determined by the maximum signal grid current for the 6k6 (10mA) and voltage?

[snowman_al]

Hi Gabe,
Before you replace the volume control you can try out adding a resistor from the ECC83 grid to ground - say 100k to start with. That does two things, it limits the volume control maximum resistance and also keeps the input grid grounded if you have a faulty wiper on the volume control. You could equally try a 47k resistor. if the noise has gone that is the problem.

Then the 220kΩ resistors are 'grid stoppers'. They are there to help prevent oscillation. That is a very high value though, normally they are 4.7k, 10k or 22kΩ.
They are on the signal grid, grid 1. That should not be passing any significant current (certainly not 10mA a 6K6 would not last long!) or have any voltage on it in your circuit. I suspect it is the Screen grid, grid 2 (that can handle 10mA) you have referenced?

Last why have you chosen 820Ω cathode resistors? You need to measure the voltage across each one. You can work out how much current each 6K6 is passing then (from Ohms law). I suspect again 400 to 500Ω would be 'normal' for a 6K6.

Well done by the way.

[Gabe001]

Chris btw your instructions for getting the 6c4 set up as a phase splitter were great. Earthing one side of the heater supply seems to have done the trick, luckily.

Snowman Al: yes I could have referenced the wrong parameter. I think 220k is too high and it could have been a mistake. I'll strap in something in parallel temporarily and see how low I can go. Re the cathode resistor, for 6k6 it would be 400R if I were using a single cathode resistor. Its 250R for 6v6. As I'm using 2 resitors, one for each cathode, it works out at 820 each. Had I chose to use the 6v6 I'd be using 2x 470-odd.

I'll try your suggestion re resistor later and report back.

After the volume pot is sorted, all that's left is the negative feedback (that'll be interesting...) and a general tidy up. Then I want to build some transmission line speakers.
https://www.ejjordan.co.uk/diy-plans/

[snowman_al]

Originally Posted by Gabe001

... Re the cathode resistor, for 6k6 it would be 400R if I were using a single cathode resistor. Its 250R for 6v6. As I'm using 2 resitors, one for each cathode, it works out at 820 each. Had I chose to use the 6v6 I'd be using 2x 470-odd...

Hi Gabe,
I think you have that back to front. The 400Ω cathode resistor in the data sheet is for one resistor (400Ω) per valve. Not a shared 400Ω resistor.
(Where 2 valves share a common cathode resistor then the resistor value is halved.)

A quick google for push pull 6K6 circuits gets you 400 or 500Ω cathode resistors.

Hope that makes sense.
Alan

[Gabe001]

I think you're right Alan. I'll replace them