Hi -- I'm trying to figure out the source of a strange hiss in amps that I have. Schematic is here.

They are AKAI/Roberts conversions (more info on the page I made here). I have received lots of help re: these amps here in the past. (Thanks!)

Executive summary: they have an EF86 stage, an 12AX7 stage, and a power tube stage (6BQ5). There is an XLR input to the EF86, an EF86 output, a 12AX7 input, a 12AX7 output, and an internal connection from the 12AX7 to the power tube with an associated speaker output.

I've replaced a bunch of caps already, to no apparent benefit. I was going to start replacing resistors, but I noticed an odd pattern when trying to isolate which stage might be causing the hiss, and so I post here. Most routes through the amp are totally quiet (obviously some noise somewhere, but basically "silent" by old tube amp standards.)

EF86-in -> EF86-out: totally quiet
12AX7-in -> 12AX7-out: totally quiet
EF86-in -> EF86-out -> 12AX7-in > 12AX7-out: totally quiet
12AX7-in -> 12AX7-out -> power tube -> speaker out: totally quiet

(for the EF86 and 12AX7 outputs, I monitor via audio interface, for the speaker out I monitor via a speaker)

It's only when I do:

EF86-in -> EF86-out -> 12AX7-in -> 12AX7-out -> power tube -> speaker out

...that I hear the hiss (in the speaker), even if the EF86 pot is turned all the way down. Simply making the "tip" connection between the EF86-out and the 12AX7-in there is hiss, regardless of the pot level. It increases with the pot up, of course, but it feels like there is a constant "floor" of hiss regardless of the pot position (the pot doesn't increase hiss until well into its travel), and only when that particular connection is made. (This isn't a subtle increase in hiss: i'm not just talking about a few dB.) The hiss happens even with nothing plugged into the EF86 input.

I have four of these amps, and all of them do it, to slightly varying degrees, but it's always the same pattern. In one amp I have replaced C25, C26, C7, C2, C10, C15, C16, C9, and C11. I have tried replacing all tubes with newly-made tubes. I have squirted contact cleaner in all tube sockets and jacks. All to no effect.

This happens whether the EF86-out in the above-described chain is from the same amp, or from any of the other amps. When using a different amp as the EF86-out source, the other amp must be turned on and warmed up for the hiss to appear.

I conclude from all this that it's likely not a defect in a particular amp but due to the nature of the circuit; but I'm not clear why that would be, since some of the routes described above would seem to have the same gain but don't exhibit the hiss, and since it seems partially independent of the EF86 pot or anything being plugged in to the EF86, and so it makes me wonder if there is some way to avoid it. Any insight? Thanks so much for your time!

Looks to me like there is 40dB or so gain from 12AX7 in the speaker out. Also probably about 40dB gain from EF86 grid to EF86 out. So, when you go EF86 -> 12AX7 -> Speaker out you have upwards of 80dB gain. It is therefore bound to hiss. Does the hiss go if you turn down VR1?

Cheers

Ian

Hi
Most, but probably not all hiss should go when you turn VR1 down.
As Ian says, both the EF86 stage and the first half of 12AX7 will give around 40 dB gain each so will add a significant amount of hiss. The pot INTO the EF86 stage will have essentially no effect on 'hiss' as all of the gain is happening after it. Having a third 'level' pot on the output of the EF stage would allow better s/n ratio in times when you don't need all the gain.
As it stands this look a 'classic' case of poor gain staging!
As an 'aside' the two 'RC combination' sections that hang off the anode of the 6BQ5 seem a little 'subtle' as they are providing some loading (minimal) of the output valve which has quite a low impedance. I presume these are 'leftovers' from it's former existence? These will not affect the hiss however.
Matt S

These guys got you in the right direction, I have that amp as well, and experienced something like what you are talking about.

Daisy chaining the two preamps Is fun, but loud.

My solution was to use a lower gain 12au7 and it helped. There's also a AT7 I think, I have a bunch of tubes that are lower gain 12AX7s... They are abundant and cheap on the bay.
I rolled them until I found what worked best.

Preamp Tube Gain Factors And Substitution Chart - 300guitars.com

Take a look at this, there's alot if info about stuff like this around.

Hey, thanks for the attention to it, everyone!

If I'm understanding you all, this means that there is a non-controlled gain happening between the 12AX7 output and the (hardwired) input to the power tube stage? And that there are also minimum gains even when the EF86 pot is all the way down (because the EF86 pot is pre-tube), so that the sum total of all these gains results in inevitable hiss? And that explains why all the other routes I mentioned are basically silent, because they aren't using all the possible of stages of gain at once.

That makes sense. I think I was basically unaware that there was gain in the power tube stage, and wasn't seeing that the pre-vs-post tube location of the EF86 pot would be important, which makes sense, too.

Not really -- it increases when the pot goes up (but only after ~50% of travel, i.e. relatively cranked), but there's a fixed floor of noise that remains when it's all the way down, as Matt said.

I should have said that only reason I chain it all together is because when using it as a guitar amp it's nice to be able to overdrive the first 12AX7 stage with the EF86, and the 12AX7 second stage with the second pot (and maybe that also overdrives the power tube, dunno?). (The XLR jack is also a 1/4" input; forgot to show that on the schematic.)

So, would it be somewhat of a waste of time to replace resistors to at least minimize this hiss?

And I'm wondering if there's any way to change the gainstaging with new pots that would reduce the hiss but still allow for the distortion. E.g. if one of the stages was just amplifying hiss but not contributing to distortion, I could pot that down when chaining it all together. I guess the obvious places would be between the EF86 and 12AX7, and between the 12AX7 and the 6BQ5?

Yes, exactly.

Thanks, yeah: so would switching to a lower-gain tube just change where the hiss appears when turning the pot, but I'd have to turn the pot up higher to get the same distortion, etc, so it'd be moot, or would it possibly help solve the issue? (I'm just not clear on the relationship between a tube's max-input-level-before-clipping and its gain.) I suppose it's an easy enough thing to just try. :-)

Maybe a main question is how much of the distortion I like is coming from the 12AX7 overloading and how much is from the power tube overloading; if the former, seems like I could gain it down with a new pot coming out of the second stage of the 12AX7 and thereby control the hiss without losing the juicy distortion?

The approximate gain of the power stage is R21/R14 which is in the region of 15dB
The gain of the 12AX7 stage before it is about 50 times or 34dB so the two together will be nearly 50dB. The gain of the EF86 stage will easily exceed 100 times or 40dB. With the 150K at the input to the EF86 turned down you will be amplifying the self noise of the EF86 by a good 90dB. Even if the tube noise was a mere 2uV, this would be about 60mV at the output which is just over -22dBu. No wonder you can hear it.

Cheers

Ian

I was going to ask "but why is it quiet when I chain the EF86 to the 12AX7 and take the output before the power stage" but then I tested more carefully than last time and... yeah, it agrees with what you're saying: same hiss. That's what was confusing me before, but I think my testing must have been a bit sloppy.

So, anything to be done if I want to chain everything and lower the hiss? If I put a 12AU7 in as johnblue suggests, will it distort at a proportionally lower input level, thus allowing distortion options with less overall gain, and therefore less hiss?

If I wanted to add a pot post-12AX7, would I just replace R16 with a 500K audio-taper?

And: no point in looking at replacing resistors, as it's likely just the tube self noise?

Hi
If I wanted to add a pot post-12AX7, would I just replace R16 with a 500K audio-taper?
You could, although the 2b valves have negative feedback which will alter the effect of the control somewhat. You might like the extra variation available.
It may actually be better to have another pot after the EF86 so you can distort that but reduce some noise.
Basically you are trading 'distortion effect' against excessive gain. There is probably nothing actually 'wrong' with the amplifier as it is, just too much gain in the wrong place for decent signal to noise.
Matt S

Hi chconner, you put a lot of work into that link and off the back of it I got myself two of these akai machines in a few to doing the mods as soon as I get round to it. Can I ask has this hum always been there since your mods and as I'm not as experienced as you and a lot of guy's out there would I be better off waiting to see if you get to updating your website with your new findings.

Cheers Louis

Hi
The reported problem is HISS and not HUM. Hiss is largely dependent on the gain and the devices used (valves are usually more 'hissy' than good transistors/ICs).
HUM is a different matter altogether and if it is HUM that you have there is a pot shown on the heater supply which can help reduce that (elimination completely is near impossible with AC fed heaters), or it could be problems with 'ripple' on the HT DC supply.

As a note, when the circuit was a tape machine the EF stage probably had feedback lowering it's gain and noise, so there would not have been nearly 90dB of gain available.
Matt S

Sorry my bad I ment hiss, I replied at work today and was in a bit of a rush. l had a tech guy build me a tube amp which I also had headroom and gain issues at first until I swapped out the EC 83's it had in to some EC 81's which made it very usable and I really like what it dose. He also put a pot which would make the gain veriable on the second valve meaning it now had a main gain and a secondary gain which only effected 1 of the tubes which helps out gain staging further.

I'm really happy with it now but there's plenty of room in there if I want to add some trannys down the line.

Also what I was referring to above is that I also came to realise that these tube curcuits are really susceptible to hiss if the gain stage was not quite right, but if chconner finds a way of combating this in some way as to make the tubes less sensitive to gain hiss then it's a job well done.

Cheers Louis

Ok! Did some experimentation with a 12AU7, as suggested.

If I do the "whole shebang" chaining, but want it "clean" with minimal hiss, the 12AU7 didn't help anything: it amplifies the inherent hiss from the EF86 much less, but it can't take nearly as much signal from it before distorting, so the EF86 has to be turned down, thus decreasing the signal/hiss ratio from the EF86. When I use a 12AX7, I can turn the 12AX7 pot down to reduce the hiss and crank the EF86 knob much higher to get to the same signal level, and the overall hiss is ~exactly the same as with the 12AU7.

(Of course there's no point in doing the chaining if I just want a clean signal; it's for the sake of utility when playing live and not having to move cables around to change distortion vs. clean, etc.)

But in terms of a distorted signal the 12AU7 made a positive difference: with the EF86 at a higher level the S/N is better and I got a bit less hum/hiss/etc overall using it (somewhere around 4-5 dB better overall S/N through the amp at comparable levels of distortion.) It's also a saner combination for chaining (i.e. the pots aren't both at like 2% all the time and hyper sensitive, etc.)

So, Matt, now that I'm understanding better your point about a new pot between the EF86 and 12AX7 (these tests totally back up your point, afaict) -- what does that look like? Replace R10 with a 500k audio-taper?

Thanks!

Honestly these amps have gone through a lot of various phases, and i've been tweaking them here and there and so forth (new capacitors, different tubes, different things fed in to them, etc) so I couldn't make a convincing case one way or the other about whether the hiss has always been around. I believe it has. And yeah, whatever I learn, if anything, I will update on the web page -- glad it was useful for you!

...and for the record, instructions on adjusting the hum controls are in the .zip file available on the web page.

Hey
If I remember correctly from the original table beast / Boris mods by Rod C, when he was daisy chaining the tubes, it was always 12ax7 into the ef86. From what I've read, it seems like you are doing it the other way around, perhaps therein lies your problem? I bet you could get a low noise 12ax7 as the first stage then on into the ef86.

Thanks! -- that makes some sense, since you have gain control in the middle of the 12AX7, which seems a little better situation than the pot before the EF86. I'll check that out for the educational value of it. But in general I need to go EF86 -> 12AX7 because I'm mostly using it as a guitar amp, and the 12AX7 -> 6BQ5 is hardwired.

I could well have it backwards in my recollection!
Are you running the ef86 heaters with the 6.3 or the 5.7 volt lines? If you don't have the 5.7v, it may be only some of the many variants that has this. I've heard the 5.7 on the ef86 will lower your noise and extend the tube life.

Interesting. From the schematic it would appear to be on the 5.7V for the EF86. Haven't measured it.

You cannot simply plug in a 12AU7 into a circuit designed for a 12AX7 and expect it to work. The two tubes are vastly different and the components in the circuit used to set the correct operating point for the 12AX7 need to be changed to get the 12AU7 operating correctly. That is why it distorts so much.

Tube rolling different versions of the same type of type number is fine, but swapping in completely different ones is a recipe for disaster.

Cheers

Ian

Hi
The best you can say about swapping between the host of dual triode types 12A (U,X,T and others and equivalents) is that they probably won't die or kill surrounding compoments. They will NOT all work 'properly' as the original designers intended but SOME permutations may present useful 'variations' of tone or distortion characteristic.
Matt S

Thanks - I was quite happy with the 12AU7 in terms of the distortion results, but my main idea is to use a 12AX7 in conjunction with Matt's recommended pot between the EF86 and the 12AX7 stage.

So, if I replace R10 with a 500k audio-taper, am I on the right track? (Schematic is in first post.)

Much obliged,
-c

Hi
It would certainly allow more possibilities in terms of desired 'distortion' and the 'hiss' due to excessive gain.
With a pot as suggested you can still have 'max gain' and at lower settings, reduced hiss.
Is this setup running into a speaker or are you feeding the input of something else? (Yet more gain!!)
Matt S

Hi
Just looked at (Preamp Tube Gain Factors And Substitution Chart - 300guitars.com).
There are several 'half truths' promoted on the first page. The valves mentioned are NOT 'all the same but with different gain' as the quoted gain figures for each type depend on the components used 'around' the valves. Another glaring 'mistake' is the maximum anode current for which a AX7 is about a tenth that of a AU7.
The gain of each type depends on the anode resistor (and further loading) and the supply voltage and the cathode bypassing (for a simple stage). If it is used with feedback then the situation changes radically.
It is about as accurate as saying 'all guitars sound the same'.
Matt S

Hey I found this related article...
Bespoke Mic Preamp based on Akai/Roberts circuit | Preservation Sound
The author suggests changing the 12ax7 stage to a redone au7 stage in order to drive a 15k:600 output transformer.

For what it's worth, I had never seen that article before, I had just studied the gain characteristics of various 12a?7 type tubes and how they can be interchangeable at times. I just googled looking for an example. I also mentioned that there is alot of information elsewhere and I was hoping to just open that door for exploration.

My experience with that amp, the Rod C. Mods, I never really opted to daisy chain the two preamps, but when I was trying to test it, I didn't like the sound, it was loud... So knowing that I could do no harm in switching out the tubes, I tried it.

I'de think in a situation like this, if the guy is going to most permanently be using this unit as a a 5watt guitar amp and always daisy chaining them, using the ef86 as the first stage providing most of the gain, it might be worth doing the homework and finding a way to use a lower gain tube for that second gain stage, my theory is that the 12ax7 has some gain that is never going to get used. At least that's how it felt when I was using it. When I had a lower gain tube in the socket, it just seemed like I had much more creamy room to play with.

I however, never used the thing as a guitar amp, only a mic amp, so I was trying to tame the distortion. I remember when I had the two Amps tied together, i couldn't push the AX7 past 1/4 before it would distort heavy. In the end, I use them separate, ef86 as a mic amp, and 12ax7 as a line amp.

Also, I think it's fair to note that this entire circuit implementation isn't / wasn't designed for the purpose anyone currently is using it for, it all used to be a tape machine.

All this talk makes me want to buy a couple more and have at it again, I would but I know I'll never beat Tablebeast, if you guys haven't checked out what he does to these and the ampex 601's you should. His work is what made me want to get these models.

Currently I have 2 ampex 600's that I'm going to hack on.

Thanks all -- I was out sick, just getting back to this:

Awesome. I'll give it a shot and report back.

Just running the speaker out into a speaker, yeah.

Success!

Finally did the operation on all four amps with great results. It knocks off maybe 3 to 6 dB of hiss, depending on the amp, and, more importantly, takes the noise well under the "too loud to be comfortably usable" threshold and into "gentle background hiss that doesn't bother you" territory.

Thanks so much to everyone that helped out!

I'd still be interested to hear any ideas for further hiss reduction (start replacing resistors?), but the amps are much more usable now. Awesome!

I even had some small original knobs left over from the conversion, and they fit right in between the larger two: http://lacinato.com/robertsmod/720A-extraknob_cc.jpg

I did forget that three of my amps have a 1M resistor to ground and a 68k resistor on the input, and one has a cap and the 500K, so I had to order some more pots. (schematic has been updated/corrected.)

Im currently parting out a junker akai m-8 if anyone needs parts.

I found this one in irreperable shape in an uncontrolled musty place. It had long been a parts picker before i got too it so its pretty well just parts now sadly.

Contact me if anybody needs parts.