Too much gain in EF86

Kinda of have to say I disagree with Gnobuddy's theory that the pentode front end was a mistake. This was the 1950's. They had vintage low output passive pickups, and no guitar pedals. They were desparate for gain, hence the use of grid leak bias in some triode input stages. Sure, the OP and others can dislike the design for valid reasons, but don't rewrite history.
 
Last edited:
They were desparate for gain, hence the use of grid leak bias in some triode input stages.
Wow, wonderful to meet a 134 year old blues guitar legend. 😀

It's absolutely amazing that you've lived to such an extraordinary age, in spite of your years of very rough living, and the time you spent in the Texas Penitentiary. 😀

Getting serious, if two identical triodes are biased to the same operating point, one with grid leak bias, the other with fixed bias with a cathode bypass cap, both will have the same voltage gain. There is no gain advantage by using grid leak bias. (That's a fact, not a guess.)

Grid leak bias was most probably used because it was simpler and cheaper. One less resistor, and no need for a cathode bypass cap. Affordable capacitors with several microfarads worth of capacitance, and low enough leakage current, were surely either nonexistent, or rare and expensive, depending on the exact time frame you're thinking of. (That's only a plausible guess, not a proven fact.)

..don't rewrite history.
Let's revisit a few public historical facts, and one fact from my personal history:
  1. Historical fact #1: VOX guitar amps with an EF86 pentode front-end were only sold for a few years, then the pentode was ripped out, and replaced by triodes.

  2. Historical fact #2: Many later copy and clone amps used the same EF86 input stage. I can't find any that stayed in production for long, either.

  3. Historical fact #3: A 'Web search will turn up many forum threads about problems with microphony in AC-15 copies, and frustrated owners who have swapped multiple expensive EF86 tubes to no avail. Here's one example, a 2012 thread on The Gear Page: https://www.thegearpage.net/board/index.php?threads/chronic-ef86-microphonics.1123412/

  4. Personal historical fact #4: I've designed and built a couple of DIY tube guitar amplifiers with pentode stages in the preamp. Having read Merlin Blencowe's book before designing them, I designed my pentode stages to have about the same voltage gain as a half-12AX7, and I never used a pentode stage right at the input. I've had no problems with microphony or excess hiss in these DIY amplifiers.

-Gnobuddy
 
So you never owned or built an original AC15 amp yourself, right? I did build one, and it sounds great and does not have microphonics issues. So far I have tried different used vintage EF86's taken from old tape recorders, I have no experience with new production tubes.
About this:
Historical fact #1: VOX guitar amps with an EF86 pentode front-end were only sold for a few years, then the pentode was ripped out, and replaced by triodes.
You are implying that was because they where always causing problems, not only in the AC30 but also in the AC10 and AC15. I think it had more to do with the fact that Dick Denney left the company in the mid sixties, and he was the one who wanted the EF86 in there.
So yes the EF86 makes the amplifier more sensitive an vulnerable, but that is the price you pay for a unique sounding amp. Let the OP suck in these opposing views and decide for himself.
 
Since this is a DIY project, might as well fit a switch to select between pentode and triode mode. I did this to the output stage of my Pignose G60, and also another switch to select between 320V and 450V B+

Is it worth it? IMO no, I always run triode/320V mode, but it was fun.
 
So you never owned or built an original AC15 amp yourself, right? I did build one, and it sounds great and does not have microphonics issues. So far I have tried different used vintage EF86's taken from old tape recorders, I have no experience with new production tubes.
I've built two guitar amps more or less based on the clean channel of the Vox AC15 because I like their clear sound a lot.

Gain in both of them is indeed high but just like gijser I don't experience microphonics/hiss/noise problems. I did pay much attention to earthing and keeping connections to the first stage as short as possible.

My amps are not combo's though, which probably reduces the chance on microphonics.

Guitar Amp EF86 + 2 x ECL86 pp.png


Guitar Amp EF86 + E92CC + 2 x EL86 pp + EZ80 + 2 x ZZ1040.png


Guitar amp EF86 + 2 x ECL86 pp

Guitar amp EF86 + E92CC + 2 x EL86 pp + EZ80 + 2 x ZZ1040
 
  • Like
Reactions: mozz
Come on, the pentode was invented in 1926 and the EF86 is from 1953 (its predecessor EF40 from 1948, EF37A from 1942).
The CF50, introduced in 1939 by Philips, should also be mentioned in this context. It has a bifiliar filament (was it the first with such a filament?) which operates at 30 V / 200 mA. It was intended for the amplification of very low level signals, much lower than the output of electric guitars.

A link to a datasheet: CF50 Philips
 
So you never owned or built an original AC15 amp yourself, right?
Correct, for the same reasons I never owned a 1960s British sports car. IMO, both are beautiful, and both are so troublesome as to not be worth the headache of ownership.
I did build one, and it sounds great and does not have microphonics issues.
Congratulations! You might want to buy a lottery ticket, with that sort of good luck. 😀
You are implying that was because they where always causing problems
Why else? It's the logical conclusion. If the products had continued to be successful and trouble-free, VOX would have continued to manufacture and sell them.
I think it had more to do with the fact that Dick Denney left the company in the mid sixties, and he was the one who wanted the EF86 in there.
So why didn't anyone else at VOX want the EF86 in there?

There had to have been a reason why the remaining powers-that-be at VOX did not want the EF86 input stage, after Denney left. What was it?

Consider: Guido Van Rossum invented the Python programming language while working for Google. He left Google after that, but Google continues to use Python heavily, long after Van Rossum left.

Google didn't drop Python the minute Rossum left. That's because Python continues to be an excellent programming language.

So why would VOX drop the EF86 front end the minute Denney left, unless they found it flawed?
So yes the EF86 makes the amplifier more sensitive an vulnerable, but that is the price you pay for a unique sounding amp.
Agree on your first point, but not on the second.

You don't have to pay a heavy price just because you want a small-signal pentode in your guitar amplifier.

It's perfectly possible to get "pentode sound" out of a guitar amplifier, without having to deal with the headaches that plagued most of those early VOX designs.

All you have to do is (a) Design your preamp pentode stage for reasonable voltage gain, and (b) Move the pentode to a later stage in the signal chain, where its higher noise figure doesn't matter.

I design my pentode stages to have roughly the same voltage gain as a typical half-12AX7 stage, i.e. about 34 dB. Usually this requires dropping the B+ voltage to the pentode to about half the triode B+, as well as making other sensible choices in operating point and anode (load) resistance.

A bonus of designing your own pentode stage is that you don't need to use pricey EF86 tubes. There are/were plenty of other NOS small signal pentodes that were never used in a popular guitar amplifier design, and therefore, escaped the silly price inflation that affects tubes like the EF86. Until a few years ago, I used to be able to buy the ones I liked for a buck apiece ($USD).

Let the OP suck in these opposing views and decide for himself.
Agree 100%. This thread has given the OP at least four different ways to lower the gain of his pentode stage:
  1. Remove the cathode bypass cap (con: may not lower gain sufficiently)
  2. Triode-strap the pentode (con: no more "pentode sound")
  3. Redesign the pentode stage for less gain (con: requires working through the design equations)
  4. Insert a trimpot between the pentode output and the rest of the amplifier, set to attenuate the excess gain and hiss (con: simple, but the EF86 will still overdrive prematurely).

-Gnobuddy
 
...introduced in 1939 by Philips...
If I ever had access to a time-travel machine, I would make a pilgrimage to the Philips facility in Eindhoven, between WWI and WWII.

Philips invented so many incredible products in that era (and for decades afterwards, too). So many have been forgotten since then.

Here is a fascinating article about yet another Philips, Endhoven, invention, a tube that quite literally helped to change world history: https://www.dos4ever.com/EF50/EF50.html

-Gnobuddy
 
...clean channel of the Vox AC15 because I like their clear sound a lot...
Here is the secret of "pentode clean tone".

It's not really a secret, of course, it's all about the high amounts of mostly second harmonic distortion you can get from a pentode.

The attached image is a scope capture of the output signal from a pentode pre-amp stage in one of the DIY amps I designed and built. I've posted it on diyAudio before.

Notice there's no "clipping" of the waveform at all. That means no harsh buzzy noises.

But there's quite a lot of asymmetry between positive and negative half-cycles. That means lots of even harmonic content, mostly 2nd and 4th harmonic.

And that's the biggest part of the "secret" of those beautiful "pentode clean tones". 😀

The other part of the "secret" is EQ. Put some treble boost in the stage(s) before the pentode. The treble boost will accentuate the harmonics from the pentode, and make them easier to hear. The sound will be brighter and more "chimey".

Don't put the pentode at the input of the amplifer, and you can have the beautiful clean tones without any of the pain.

-Gnobuddy
 

Attachments

  • DSC_0562C_889_x_500.png
    DSC_0562C_889_x_500.png
    251.2 KB · Views: 73
Let's revisit a few public historical facts, and one fact from my personal history:
  1. Historical fact #1: VOX guitar amps with an EF86 pentode front-end were only sold for a few years, then the pentode was ripped out, and replaced by triodes.
I would like to know where you got this from, I suspect it sprung from your own imagination. At any rate, it is not true.
The AC15 (and AC4, AC10) amps with the EF86 were built and sold by JMI under the VOX brand for 10 years, from 1958 until 1968. In 1968 the mother company went bankrupt (wich was not the fault of VOX) and that was the end of the JMI era. No pentodes were ripped out and replaced by triodes, production of the AC15 and other amps just stopped. The AC30 production was soon restarted (because it was the most popular and profitable model?), along with newer solid state amps.
 
He wrote how many words making fun of my alias?
Oh my word. 🙄

Just a little light hearted fun about a legendary guitar hero. I'm not attacking you in any way - why would I, when I don't even know you? And you know you're not actually the real leadbelly, right? If I joke about a total stranger who died 73 years ago, how on earth does that insult you?

And how exactly is it a failing of mine that I taught myself how to touch-type? Who decided that an Internet post on a technical forum should never be longer than five words plus an emoji of a cat licking itself?

We discuss complicated technical issues on this forum, and that requires more than a few words to address. You don't have to read them. Do us both a favour, and put me on your ignore list.

he's a master of argumentum ad nauseam
Utter nonsense. You seem to be confusing facts with arguments. I've presented plenty of facts, and have been careful to be very clear when I presented an opinion rather than a fact.

Watch the video of the guitar tech swapping one EF86 after another into an amplifier, only to find every one is microphonic. That's not argument, that's fact.

Read the Gear Page thread I linked, which includes multiple posts from upset and frustrated owners of AC15s and clones, who experienced unacceptable problems with microphony. That's not argument, that's fact.

A quick search on You Tube will find other videos from amp techs describing the problems they encounter with microphonic EF86 stages in guitar amps. One guy says he's had to swap "a bucketful" of tubes in the search for a non microphonic one:
This is what Philips advised about the ECC83 versus the EF86 in low-level applications (Source: Page 23, Hi-Fi Amplifier Circuits, E. Rodenhuis, Philips, 1965):
Thank you, Kwadjo, for posting that (I've seen it before, but it's very relevant).

Philips researchers and engineers of the time seem to have been among the best in the world, so I"m not surprised that they specified suggested EF86 usage very thoroughly.

So, from that document, Philips recommended usage with a sensitivity of "50 mV for 5 W output". It's a bit of a mixed specification (voltage at the input, power at the output), but once the speaker impedance is chosen, it's basically a specification of the maximum voltage gain that's acceptable (voltage gain from pentode input to loudspeaker terminals).

A reasonable question to ask, then, is "Does the AC15 meet Philip's specification?"

I don't have an AC15 to measure. Those of you who do, it would be wonderful if one of you would make an end-to-end voltage gain measurement, with all knobs set in typical positions you'd use at a loud live gig, and ideally, a second measurement with gain maxed out, so we know the full end-to-end voltage gain of the amplifier.

You'll need to use a dummy load in place of the speaker, to avoid microphony and painfully high SPL. You'll also need to attenuate the input sinewave down to very small amplitude to keep the amplifier from overdriving. If it overdrives, the measurements will be invalid, as they'll produce a false (lowered) voltage gain. My suggestion is to start with no more than 1mV pp at the input, and lower it further if you see clipping distortion in the output signal; you may need to knock up a two-resistor attenuator to knock down the signal, if your signal generator can't be dialed down all the way to 1 mV pp. Many can't.

Even though I don't have access to an AC15 to mesure, we do have enough information to make a solid estimate. Let's do that.

Let's start with the typical output voltage from guitar pickups. Rod Elliott has a characteristically thorough table of guitar pickup voltages he measured on his website, here: https://sound-au.com/articles/guitar-voltage.htm

The key takeaway is that almost every voltage Elliott measured is less than 50 mV RMS. Many are under 20 mV RMS. Measurements on thinner strings are even smaller, around 12 mV RMS.

So now we know for sure that a hypothetical guitar amplifier that just barely reaches full output power with an input of 50 mV RMS will produce only "clean tone", i.e. no overdrive at all. In fact, it probably will never reach full output power at all.

A successful amplifier designed for clean tones has to have more sensitivity than that, to ensure you can get the full output power from it. For instance, Fender Corp specified a "30 mVAC" input signal for testing the '65 Princeton Reverb (see attached image). This amplifier was designed to produce only clean tones - overdriven guitar wasn't on Leonidas' mind at that time and place.

The Fender schematic doesn't specify if the "30 mV" was actually RMS, peak, or peak-to-peak. If we assume that an oscilloscope was intended to be used for the measurement, "30 mV" most likely meant "30 mV peak to peak", as peak-to-peak is easiest to read off a 'scope screen.

30 mV peak-to-peak is a little less than 11 mV RMS, for reference. This is in good agreement with Rod Elliott's measurement of 12 mV RMS from the open 1st string E of two different guitars he tested.

Okay. So Fender thinks an input sensitivity of 30 mV pp is about right for full-power, unclipped output from an amplifier designed to produce only clean tone (no overdrive).

In order to overdrive, a guitar amplifier has to have much more gain than that. How much more? That depends on the degree of overdrive wanted. Gain needs to be at least ten times more (another 20 dB) to get any significant overdrive. That means an amplifier that starts to clip with an input of 3 mVpp (ten times smaller input signal than 30 mV pp).

The VOX AC15 was famous for overdriving easily, so it's likely that it's input sensitivity for full output was even less than this. But let's make the conservative assumption that it took a full 3 mVpp to drive it to full output.

The amplifier was rated to produce 15 watts RMS, so let's use that number.

It takes almost exactly 31 volts peak-to-peak to produce 15 watts into an 8 ohm speaker. We've estimated the AC15 was designed to reach this full output with an input signal of 3 mV pp.

Therefore the voltage gain of the pentode channel of an AC15 can be estimated at (31000/3), or 10,328. (That's +80.28 dB, if you prefer).

Phillips specified a gain of no more than necessary to produce 5 watts RMS at the output, for 50 mV at the input.

Unfortunately, the little excerpt from the Phillips document doesn't specify if this is 50 mV RMS, 50 mV peak, or 50 mV peak-to-peak.

Let's assume peak-to-peak, measured on an oscilloscope.

With a gain of 10,328, an input signal of 50 mV pp would produce an output signal of 516.4 volts peak to peak. Obviously an AC 15 can't actually put out this much voltage - but this calculation tells us if the AC15 meets the Philips specification, or not.

516.4V RMS output would produce 4166 .7 watts, RMS into an 8 ohm load.

So: we've estimated that the AC15 has so much gain that it would try to deliver over 4000 watts RMS to the speaker for a 50mV input.

Philips specified that their EF86 should be used in an amplifier that does not deliver any more than 5 watts RMS to the speaker for a 50 mV input.

In short: our estimate says that the AC15 violates the Philips EF86 specification by about a factor of eight hundred.

If that's not an engineering design mistake, I don't know what is.

Again, if you disagree vehemently with my rough calculation above, please do the measurement yourself.

(Remember: 15 W RMS means 31 Vpp at the speaker. With an estimated voltage gain over 10,000, that means the input test sinewave needs to be less than 3mVpp to produce a valid measurement. The actual amplifier gain may be higher, I don't know; you shouldn't feed more than 1 mV pp, sinewave, into the input, if you want to measure an accurate voltage gain for the entire amplifier.)

-Gnobuddy
 

Attachments

  • Princeton_Reverb_Input_Stage_Schematic.gif
    Princeton_Reverb_Input_Stage_Schematic.gif
    40.8 KB · Views: 103
You wrote: "Philips specified that their EF86 should be used in an amplifier that does not deliver any more than 5 watts RMS to the speaker for a 50 mV input.". But I think you are wrong about this. The number of 50 mV at 5 Watt (and a loudspeaker of 5% efficiency) is applicable to the ECC83, not to the EF86.

On page 4 of this Philips datasheet for the EF86 it says the correct number is 5 mV for an output of 5 Watt: EF86 Philips

Besides that: Assuming your calculations are right, don't they show that every guitar combo which uses an ECC83 in the first stage is an 'engineering design mistake'? And that using an EF86 instead of an ECC83 would be at least less of an 'engineer design mistake'? Would it even be possible according to your calculations/verdict to design a combo in that era that would not be an 'engineering design mistake' in the first place?