Valve replacement - matching

[Edwin]

At present I don't need to face this problem, but the time may come. I have the WD88VA with four KT88 valves. If one of them failed, I assume I would have to replace it and its corresponding valve in the other channel with a matched pair. But that wouldn't account for the difference in match between the remaining two, and the one that has been taken out to allow for the new matched pair might be a better match with one of them. The ideal solution would, I suppose, be to replace all four with a "matched quad". You'll see that I'm no expert, and I've been reading all sorts of warnings about valves sold as "matched" pairs or quads that don't live up to the description. Any advice that I can put in storage for the future?

[pre65]

Ideally a matched quad (4), but if using matched pairs use them both in one channel.

[John Caswell]

Hi Edwin,
There seems to be some confusion as to where matched pairs/quads of valves should be fitted.
In the WD88VA for example a matched pair should be fitted in one channel ie left or right and not separated one to the left and one to the right as you suggested.
If you have a matched quad then theoretically you can place them at random in either channel as they are a "matched quad".
Practicall, however, there will always be minor differences between the valves, so what I do is initially fit the quads, then measure the cathode voltages and swap the valves around so that those with very similar cathode voltages are on the same channel. This at least will mean that the anode currents are fairly similar through the OPTx so should minimise the induced HT rail hum through the OPTx.
If you look at the circuit of the original Williamson amplifier you will see that he took a lot of trouble to provide adjustments to take care of slight differences in valves, even though he worked for GEC/Osram so had access to many many valves to try

John

[bikerhifinut]

I guess that for what it costs, you can buy matched pairs for each channel.
However, bear in mind that WD didnt supply matched valves with the kit in the first place. It's less of an issue with the cathode bias design than with fixed bias where matched Valves are important.
(Fixed bias can be confusing as thats the sort you have to adjust fairly frequently as the valves age)
All I ever did was match the 4 KT88's by cathode voltage and got them near enough.
Hope this helps.

[Richard]

Even cathode bias amps should have channel pairs matched within 10% current and preferably similar slope.

WD valves may not be marked but will have been supplied to a reasonably tight tolerance. EI told me that Maplins, for example, weren't marked but were graded within 10% for all their shipment. I bought 5 odd el84 from 2 stores here and sure enough they were all close.

Don't take this for granted though with other suppliers. Just taking bulk stock can lead to wild variations. I tested 22 bulk Svet 6550C and made 3 good quads from them but they varied something like 48mA to 70mA and could have not been used in p-p without some form of grading.

If you don't have a tester best way is to check cathode volts (pin 8 to ground) and compare, using similar ones in a channel pair. (Divide the cathode volts by the cathode resistor ohms to find the current through the valve).

Then you can go a step further and switch them within the pair and listen for less hum. Bear in mind the valves themselves might be spot on but there can be differences in the TX windings, tappings and grid resistors etc Really you're just wanting to be sure they're not humming which indicates one is drawing more current than the other and can lead to runaway and overheating in that valve.

All this is usually only an issue with a valve which fails early on. If they have 2000+ hours on them most folk buy a new matched Quad same or a different flavour and keep the old ones for spares. If one has failed early it might be worth sending the pair to a decent dealer with a tester who sells that brand such as Watford, Langrex etc and having them match one to it which isn't difficult.

[Edwin]

Thanks to all of you for (very tactfully) putting me straight on several basic points - mainly that I had totally misunderstood "matching" to refer to the matching of the left and right channels. If you replace both KT88s in one channel because one has failed but leave the elderly two that are still working on the other channel, is it likely that you would notice some audible difference between the two channels?

At present, all's well and there's no hum even at full volume, so I guess from what Richard and John say that the match of the existing valves is good.

[John Caswell]

Hi Edwin,
"If you replace both KT88s in one channel because one has failed but leave the elderly two that are still working on the other channel, is it likely that you would notice some audible difference between the two channels?"
I think that will depend very much on the individual and also how his/her brain functions. It goes back to perceived differences.
As an example, I repaired an amp and replaced the same given resistor in both channels. The customer was convinced that the faulty channel was different to the working one, as I hadn't told him (deliberately) that I had replaced the same resistor in each channel. The moment I pointed that fact out the problem went away. The moment you think there is a problem you are going along the slippery path. When people do an upgrade and say the item is now "better" I ask them to define "better". Quite often it is "different"
I suspect in your case the most likely thing you may notice is a difference in gain, but, unless severe will will not be apparent on most things, however constant tone/measurement will show it up.

John

[bikerhifinut]

All very educational.

I've never really understood how to quantify valve life on power valves.
I'm sure my KT88's are well past their nominal service life, but they still measure ok, and sounds fine although i may not have noticed subtle changes in tonality over time.

I guess that the WD88VA runs the KT88's fairly conservatively, lowish HT and modest power output, so they'll last well?

My take on matching came from sales blurb for the original Leak power amps, claiming that with automatic bias you didn't need to buy a pair if an EL84 went. I suppose back in the day a couple of quid for a valve represented a fair old chunk of the average weekly wage.

Interesting about Maplins valves. I bought 4 not long before they stopped doing them and they paired up well on their matching batch numbers. Incidentally they were 6P14P-EV russian jobs from the reflektor factory, with an Edicron branding. I think they sound pretty bloody good in the old stereo20.

[Richard]

Hi Andy,

Yes the Maplin EI EL84s used to be Yugo, almost a dead copy of Mullards, which sounded great when I bought them for my ST20 too.

You're probably right about premium quality valves from the likes of Mullard and GEC; they may well have been to a tighter tolerance for the retail market. I don't know though, I can only remember back to the 60's when a local "serious" radio shop had an Avo tester on the counter so maybe they matched them even back then.

I don't have the voltages to hand but seem to remember WD88 runs them quite hard with fairly low voltage buit highish current to hold the stage in class A iirc. You can work out the plate disippation but around 3000 hours will be safe.

I agree with John that one might notice a difference with a new and an old pair. New op valves sound bright for around 200 hours then mellow for 2-3000 hours then go off and start sounding weak with less bass and treble. At that stage either the user notices and changes them or they continue to lose emmission and pass less current. One usually goes before the other in a pair and the imbalance causes first hum then red plate on one valve then thermal runaway and catastrophic failure.

On the other hand the acoustics in a room and the channel signals are asymmetric anyway so you might not notice age or brand across the channels