I have read your FAQ with great interest, learning a lot and answering questions I have had for years. It has reminded me though of two points that I really want to understand and which I cannot find straight answers to. Please could you help?
The first is why I can't simply take two of the 4 EL34 valves out of my 100 watt amp to get it to run at lower power? The valves seem to be in parallel pairs and so surely taking out one from each pair would just cut the power getting through to the output transformer?? I'd love to be able to do this but Marshall warn against doing this but without saying why (just that it will damage the remaining valves/transformer but not why)
The second query is why switching a valve amp on with no load is so bad for it. OK I have lived for years just accepting I shouldn't do it but the kid inside me wants to know why.
In my experience the kid in you is way more likely to run amps with no speakers plugged in than is the intelligent questioning adult. If I plotted proportion of jobs featuring burned output transformers against the average age of clients I think we would have an indicative curve.
Both of these are output impedance matching issues, so I have to explain that first before moving on to specifics.
Impedance is resistance to AC. It's measured in ohms, the same as resistance, but you can't read it on an ohmmeter.*
Your output valves will work best if they are looking into (or driving, or loaded by, pick your metaphor) a certain output (or 'plate') impedance range. This will be a few thousand ohms per valve; each valve type has different expectations, and the type of circuit changes impedance expectations too. Your output transformer is an impedance converter (amongst other things) - it multiplies your speaker impedance into something your output valves want to see.
A typical output transformer will turn an 8 ohm speaker load into, say, 8000 ohms for the valves - in a typical push-pull amp with paired output valves, 4000 ohms per side.
Valve output impedance is a ball-park affair. A given set of valves might be happy looking into anything from (say) 3000 to 8000 ohms. So plugging 16 ohm speakers into an 8 ohm load is probably not too scary.
Right now let's deal with pulling two valves out. This doubles the output impedance that the amp requires from its speaker (equal impedances in parallel, like the two valves one one side of a push-pull quartet, produce half the impedance of each component). So to match the doubled impedance requirement from the valves, we should double the speaker impedance. Or if possible turn the selector down one click. So... with two output valves pulled out playing into 8 ohm speakers we should have the selector on 4 ohm - or arrange to have a 16 ohm speaker loading. But if you don't do that it will probably still be ok, output impedances being a ball-park business.
Why do Marshall say don't do it? Because for all they know you are 17 (Sorry 17 yr olds but statistics are statistics).
Why do output valves need a certain output impedance range? Because that impedance will also set a voltage/current relationship that they can handle. V=IR, or voltage = current x resistance, as every 17 yr old once knew and has now forgotten anyway.
OK now for running valve amps without a speaker. With no speaker plugged in there is, in the typically-wired amp, no load at all, which is an infinite impedance. Put an infinite value for R into V=IR, remembering that the bias and the potential in the amp will try to hold I constant, and have a think about it. What happens is that the valves make attempts to chuck out enough voltage to drive an infinite impedance. They can't, of course, but they are high-voltage devices and they have a damn good try. The flash voltages they generate can be high enough to cause arcing inside your output transformer- and outside it too, spectacularly. Arcs can occur inside the valves too, and on the valve bases. Internal feedback from these events probably makes things even worse. Typically one or more valves go out, and there are fused turns in the output transformer, causing either no output or reduced, distorted output.
There you are. Any questions (or indeed corrections, I'm a repairman not a scientist remember) - use the comment button.
*Indeed you can't measure impedance on an ohmmeter, but if you want to know the impedance of a speaker or combination of speakers, an ohmmeter across the terminals or the jack will read approximately 70% of the impedance in ohms. So an 8 ohm speaker will read 5 or 6 ohms.
Hi Steve, I read all your post and I found it really interesting and I love the humor you use in.
ReplyDeleteI'm two time 17 but need some confirmations.. a couple of question.
I'm owning a SoundCity 120, as you can immage is a little to loud to properly work, its have 6 output tubes. Based on your article seems is not a problem if I pull out a pair of them. Wanna say, if I've a 16ohm speaker load, amp selector on 16 still will be ok?
Also, I own a SoundCity Concord.. (yep, love SC amps..). Its output transformer have just a 8ohms secondary, but my configuration have two internal 8ohms speakers in series, so 16 ohms. It has been modified to have an external speaker wich doesn't exclude the internal one if connected (and with a 16 one in parallel all seems ok).
Without the external speaker will the amp suffers the situation (OT 8ohm - SPK 16ohm)? I mean, if V=IR and intensity tends to be constant, raising resistence will duplicate tension, is this fine?
hope my english is.. enough english, many thanks.
Mauro
Hi Mauro. Your English is plenty English enough! - and thank you for taking an interest in my blog.
ReplyDeleteYour 120 will be fine with a couple of valves pulled. As you know, to compensate with the impedance selector you would have to have it on a higher impedance than the cab's, which you can't, but we are not talking here about a big difference.
The Concord will I think be OK too - an 8 ohm output looking into 16 ohms is not a serious mismatch, though I would be concerned about 4 ohms into 16.
Actually the impedance of speakers varies with the frequency - they show much more impedance to high frequencies than to low, for instance. This complicates the whole thing.
It's not really a simple V=IR calculation (despite what I said!). The AC voltage across the OT is limited in its 'swing' by what the power supply can deliver, though it may increase some to match a higher impedance. AC current will also try to compensate, but it is limited by the bias on the grids. Remember that valves themselves have an internal impedance, roughly equivalent to the plate impedance, but variable. Faced with an infinite impedance, or a very high one, things can go unstable, but otherwise valves are forgiving beasts and do compensate for a certain amount of mismatching.
...should have added this: the damage caused to a valve amp with no speaker connected is actually caused by the output transformer feeding back on itself in response to rapid signal changes to create spikes of very high voltage. These will arc across inside valves and harm them, or across windings/terminals on the OT and burn it. So like I said, not actually a simple V=IR calculation, though it is true in a sense that the amp is trying to generate very high voltages to cope with an infinite load.
ReplyDeleteSteve, thanks a lot, you clarified some things to me. A little step for the humans but a big one for me! :)
ReplyDeletemauro
p.s. just joined JJ & the Jukes on the space :)
...well bless your heart
ReplyDelete