Circuit Digestion - Ampex MX-10

[RicharD]

This is not a geetar amp but actually an old 4 x 2 mixer made by Ampex.  It used EF86 and 12AU7 vacuum tubes.  I've serviced 2 of these over the years (usual stuff, cap job, replace the old diodes with 1N4007's).  One of the owners swears it's the greatest tube mic preamp ever.  The general consensus on the interwebs is, "meh... OK on snare drum, cruddy input transformer."  None the less, it has some interesting things happening in the wiring.  I thought it might be a fun circuit to pick apart.  So......

The mic input transformers were made by Beyer Dynamic, a German company if I'm not mistaken.  Ampex was founded on recording technology swiped from Germany at the end of WWII.  Also the EF86 is not an American tube.  I found this odd whereas everyting Ampex I own is USA through -n- through.

The first tube stage is a EF86 wired in triode mode.  My guess is they ran with triode mode because it has less gain and therefore less distortion.  Seems like an odd choice, but the whole thang is odd and that's why I'm typing about it.  From here it hits a 100k level pot, a left right switch, & a summing resistor before hitting the next EF86 gain stage.  This stage is wired pentode mode and sets the bias of the following 12AU7 cathode follower stage. It's screen voltage is derived from the cathode of the 12AU7.  The cathode of the 12AU7 is grounded via the shield of the EF86.  I assume this was done for some sort of noise rejection, but it seems like it should add noise.  There's a touch over 6mA of current flowing through the shield.  I'm sure this is some old trick of the trade I'm not aware of but I'd like to understand the "why" this was done.  There is an adjustable feedback network tapped from the 12AU7 output and fed back into the 2nd EF86 stage.  They made this adjustable so the stereo pair could be balanced.  Finally there is an output level pot connected directly to pin 3 of an XLR jack.  Huh... pin 3???  I thought when using XLR for hiZ you used 2 hot and grounded pin 3 along with pin 1 shield.  I've got a pair of XLR to RCA cables I bought decades ago and they're tip to pin 2.  Yet another oddity.

The B+ power supply is is a voltage doubler fed from a 117VAC secondary.  The caps seems small for the job, but electrolytics were expensive back then, but this is Ampex not Silvertone.  I'm cool with voltage doublers.  You get full wave rectification and my redraw uses an "off the shelf @ Mouser" transformer into a voltage doubler with larger cap values.  There is a RC dropping node, then a "T" to 2 more dropping resistors for the left and right channels, then another pair of dropping resistors that come back to a common point for all the input tubes.  That's different.  I understand the "T" to left and right channel supplies, but what I don't get is why it "T's" back together for the input tubes.  We're not talking about a whole lotta current and the only reason I can think of for a parallel path is load sharing.  The filaments are 12VDC full wave rectification.  The EF86's are heated in series and the 12AU7 is heated @ 12V with pin 9 grounded.  This seems like a good way to do it.  My redraw uses my typical 12V regulated filament supply that I use all the time.

I think I pretty much understand most of the circuit.  I'm curious about the cathode follower being grounded through the shield of the previous stage.  The output going to pin 3 doesn't seem right.  I'm really surprised there isn't a balanced output option.  The parallel power supply node stumps me too.

Attached are copies of the original schematic, and my redraw.

-Richard

OOPS!
The original MX-10 Ripoff schematic had a typo.  the 12AT7 is supposed to read 12AU7.  Attachment has been corrected & replace.

[HotBluePlates]

Also the EF86 is not an American tube.

Do you have any date codes off of parts in the actual mixer? I don't know exactly when the 5879 and the EF86 were introduced; maybe the EF86 came out first. Both were advertised and sold as low noise pentodes.

The first tube stage is a EF86 wired in triode mode.  My guess is they ran with triode mode because it has less gain and therefore less distortion.  Seems like an odd choice, but the whole thang is odd and that's why I'm typing about it.

The EF86 is low-noise by pentode standards. All tubes have a couple of mechanisms by which they produce self-noise. All pentodes have 1 extra mechanism compared to triodes, so all pentodes have more noise than an equivalent triode. So you take a low-noise pentode, strap it as triode, and further reduce the noise coming from it.

That's dead-common in hi-fi and pro-audio.

The real issue is that the designer looks at how much noise is produced by the source. Then they usually use an input transformer to step up the signal level ("free" gain) before hitting the first tube grid. If the transformer ratio is high enough and the first stage self noise is low enough, the noise coming from the source (mic, line, etc) is higher than the self noise of the input stage. That makes the noise contribution of that first stage irrelevant.

You might consider measuring the step-up ratio of one of the input transformers.

That step-up is likely a needed item anyway, to better match the input impedance to the mic impedance. If I remember correctly, the input impedance for this mixer will look like the 22k on the EF86 grid divided down by the step-down ratio looking back to the mic. EDIT: That might be only if the resistor spanned both output leads for the transformer; as shown, the 22k is probably just a grid stopper.

Anyway, if you get some gain from the input transformer, you don't need as much from the input tube, so triode operation again works out.

From here it hits ... a left right switch

Left, center, right? That's what it looks like, and would be typical of old-school mixing.

This stage is wired pentode mode and sets the bias of the following 12AU7 cathode follower stage. It's screen voltage is derived from the cathode of the 12AU7.

Well, I would have described it as EF86 direct-coupled to a cathode-follower, but, yeah. The screen voltage coming from the cathode follower looks like a clever way of having a free d.c. supply at a new voltage level. You'd really have to think the plan out to make all the voltages between these stages work out properly.

It would be interesting to see if the current for a given signal in the EF86 is approximately equal to the current in the 12AU7 cathode follower. If they are the same or close, the load on the power supply is lightened, and it would take less filtering to decouple these stages from the input stages. Obviously, the d.c. drawn by the 12AU7 is much, much higher.

The cathode of the 12AU7 is grounded via the shield of the EF86.  I assume this was done for some sort of noise rejection ...

Do you have pics of the inside of this mixer? Are the sockets very close to each other? Is the wiring style "point-to-point" as in the resistors span from one socket to the other?

If so, there is no noise rejection going on. The shield is tied to more than 1 pin, and one of the pins is grounded. That makes every other shield pin a convenient ground lug. If the sockets are close together, and the resistors make the connection, then you have very small wiring loops for this part of the circuit.

The cardinal rule with layout and lead dress (especially in radio) is that you want to minimize the loop formed by the circuit wiring. The smaller the loop. the smaller the "aerial" you have created to pick up hum and interference. This odd connection may be only about making the shortest, easiest wiring possible.

I don't think it is about noise rejection, other than preventing noise pickup. That's because once this tube element is at ground potential, how can/does it interact with the plate? It's not heated, so there shouldn't be any real current flow between the two. The only effect I can see happening is that which the shield would do anyway: change the amount of capacitance from plate to ground.

If there is a strange uncommon trick being used, maybe PRR knows it. But I think it's simpler than it looks.

[HotBluePlates]

There is an adjustable feedback network tapped from the 12AU7 output and fed back into the 2nd EF86 stage.  They made this adjustable so the stereo pair could be balanced.

Is that control internal only? I'm guessing it would be a front-panel adjustment, maybe by screwdriver only. It's likely to set level, as it says, but when both channels are properly set, they're in balance. I'm thinking if this were part of a tube console, you'd want to be able to adjust to have a known output with a known input. After all, you have volume controls on each of your source preamps to set the left/right balance.

Anyway, there's a famous german mic preamp that uses feedback to set the preamp gain. Here that's no so much the case. But we have another thread going talking about tubes drifting. Feedback minimizes the effects of tube drift and the change of gain as tubes age. A good bit of the EF86's gain is being thrown away in feedback, probably to minimize the effects of aging, along with giving an easy level-set facility.

Huh... pin 3???  I thought when using XLR for hiZ you used 2 hot and grounded pin 3 along with pin 1 shield.

Think "balanced" and "unbalanced".If you look back at the input from the mic, you'll see the shield is on pin 1, and 2 and 3 are used to feed the input transformer.

You're right, the most common standard in the U.S. is pin 2 hot, with pin 3 as neutral and pin 1 as shield. However, there was also a "standard" with pin 3 hot and pin 2 neutral. I want to say I remember that as being typical of japanese equipment, but maybe others as well. Regardless, if you want to wire for pin 2 hot, go for it.

The caps seems small for the job, but electrolytics were expensive back then, but this is Ampex not Silvertone.

40uF is small? Where's the output tube stage pulling big current swings from the power supply? The preamp EF86's draw less than 2mA of plate current, so they can't be swinging more than 2-3mA of current each with signal, probably much less. The output EF86 is only drawing ~0.5mA for the plate, so its current swing with signal must be tiny. Worst-case, we're still only talking around 10mA of varying current for the entire mixer. That's very easy to filter.

I understand the "T" to left and right channel supplies, but what I don't get is why it "T's" back together for the input tubes.  We're not talking about a whole lotta current and the only reason I can think of for a parallel path is load sharing.

Sounds like a plan to me. The stock plan is smarter than I probably would have done it. You have to drop from 300v to 260v, and that 260v needs to feed a left and right channel that should be decoupled from each other. 2 spurs from the common 300v point looks like a good way to do it.

But then you also have to drop from 260v to 243v. If you simply drop this from 1 of your 260v nodes, you increase the current through that branch, and upset the apple cart. By pulling the current through both 260v spurs equally, you drag everything down by the same amount. Clever. Note also that each input preamp stage has its own decoupling resistor and cap. So everything is still isolated. Someone put a good deal of time into this circuit thinking out each part.

[HotBluePlates]

I'm really surprised there isn't a balanced output option.

I'm guessing this whole thing is either rackmount or intended to be part of a recording installation. If so, the output from this mixer would have gone to some kind of master L-R buss or to tape input or cutting lathe input. If so, you're not going to run the wiring a long distance, and balanced operation isn't needed.

I suppose you could add an output transformer to balance the signal if you'd like. That would be the way to do it, since the circuit doesn't use push-pull to provide an automatic balanced output signal. The added transformers will add significantly to the cost.

I'd also suggest a deeper investigation before switching the cathode follower to a 12AT7 as you have shown in the "rip-off drawing". I'd like to know for sure if there was an important reason for using the 12AU7. It might be because of the lower native internal impedance and higher Gm, which means a lower output impedance than a 12T7 when both are used in a cathode follower configuration.

That may also be your answer about why no balanced output. Maybe for the intended short run of output cabling and the low output impedance from the cathode follower (you know from the schematic it's below 25k), balanced output wasn't deemed necessary for low-noise. Or the intended next step in the signal chain didn't have a balanced input.

The EF86's are heated in series and the 12AU7 is heated @ 12V with pin 9 grounded.  This seems like a good way to do it.

Funny! The 12AU7 is the d.c. reference for the whole heater circuit...

PRR may have some corrections for things I overlooked, or guessed wrong about.

Thanks for the cool circuit and the thought you already put into it. I think you already had the circuit sussed, you just had a couple stumbling-blocks.

[RicharD]

Thank you HBP.  That clears up quite a bit.  The 12AT7 is a typo.  Dangit!  It is supposed to be 12AU7 like the source drawing.  Me and my chubby little finger.......

I didn't think to take any gut shots and it's been a while since I recapped one of these.  There are some gut shots of the MX-35 (same animal different box -n- knobs) but they've all been recapped and nothing revealing about the cathode follower grounding through the EF86 shield.  It is wired point to point and most of the resistors are carbon comp IIRC.  There are 2 resistor notes.  The plate resistors for the 1st stages are called out as "film" (I assume carbon film given the era of the unit) and the cathode resistors for the 2nd stages are called out as wire wound.  Film resistors on the plates makes sense for noise reasons.  I'm guessing the WW of the day were more precise (probably still are) and it would seem important for the 2nd stage to be very stable since that where the feedback balance happens.... and yes they are screw set potentiometers in the FB network.

>Funny! The 12AU7 is the d.c. reference for the whole heater circuit...

I noticed that which got me thinking about my filament supply being grounded on 1 side.  1 pair of the EF86 filaments don't have a ground reference.  I keep a copy of that supply built on my protoboard and use it with 12A*7 tubes all the time, but I have not used it with stacked 6V tubes.  I don't think it's a problem, but I don't think it's necessarily good form either.  Fixed 6V regulators of sufficient current aren't exactly common but variable regulators are.  Regulation for filaments might be considered overkill, but as I delve into hi end tube audio, overkill has it's benefits.  This brings me back to my comment about the B+ filter caps.  It is not at all uncommon to see parallel 470uF B+ filtering in modern "studio grade" tube circuits, ones drawing much less current.  Of course most of the "tube" stuff has a single 12AX7 loafing along in front of a whole string of FETs and/or opamps.  I compared simulations (psud2) of my PS using 100uF caps to the original circuit using 40uF caps.  Neither one displayed any ripple.  The only difference is mine charges about 4 times slower which may or may not be beneficial with regards to cathode stripping while the tubes warm up.  Suffice to say 40uF is more than adequate and once again I am blinded by bigger is better.

[HotBluePlates]

There are 2 resistor notes.  The plate resistors for the 1st stages are called out as "film" (I assume carbon film given the era of the unit) and the cathode resistors for the 2nd stages are called out as wire wound.

Well, carbon comp have the most noise, and resistor noise increases for higher resistances. Film resistors are lower noise, so it makes sense that they would spec a carbon film resistor for the plate load of the input stages.

Wirewound resistors have the least noise of all, but it is very hard to find them in high values. When you do, you might have to worry about inductance (but not so much in audio). It looks like if you opted not to use the XLR, the terminal strip uses the top of the wirewound resistor as the negative side of the signal. Maybe they wanted a special low-noise resistor for that position.

I noticed something else that should have been obvious before. The call out numbers in circles are d.c. voltages, but the ones in boxes are signal level in dB. I don't know what reference level they're using, but... The mic level is given as -60dB. Notice that after the input transformer, the level at the input grid is -37dB. You already got a 23dB boost just from the input transformer. The triode-mode EF86 then bumps it up to -10dB. Almost as much gain from the input step-up as from the input stage.

The final output level is +12dB, so there's a 40dB boost from the pentode EF86 stage (plus whatever loss the cathode follower has). I'll admit the screen resistor connection was just messing with my mind. Local feedback connection? Naw... the screen bypass cap keeps that from happening.

If you knew the reference level for all this, you could use a signal generator set to provide the proper input level, and a meter that reads in dB to check each test point, and verify proper level throughout the mixer. In fact, if the reference level is 1mW in 600 ohms, then the handy dandy HP 400D AC meter has a scale and range settings calibrated in dB. The number would read directly off the schematic and match the meter face. +12dBin that case seems pretty hot, unless you're feeding faders or equipment that needs some serious spank.

I noticed that which got me thinking about my filament supply being grounded on 1 side.  1 pair of the EF86 filaments don't have a ground reference.

I'm not 100%, but I think I've seen a whole regulated heater supply floated, and the artificial CT resistor used at the output cap to balance the circuit to ground.

And what pair of EF86's are you talking about? If you mean in your rip-off circuit, I though a pair of EF86's were omitted because the circuit is being used as a stereo mic pre.

It is not at all uncommon to see parallel 470uF B+ filtering in modern "studio grade" tube circuits, ones drawing much less current.  Of course most of the "tube" stuff has a single 12AX7 loafing along in front of a whole string of FETs and/or opamps.

Maybe the designers are too used to solid state, where all the caps are enormous. Or maybe they all go for the overbuilt regulated power supply and big filter caps. Or maybe they are sitting on a 10,000-piece lot of 470uF caps for all their other products...

I've got some cool mic preamps that a knowledgable guy made for me quite a while back. There is a seperate power supply box with an umbilical to feed B+ and d.c. heater power to the preamp. Both supplies are off-the-shelf regulated supplies, and he simply did local decoupling and filter caps to have the filter right at each stage. I don't recall, but the local caps were probably 10-22uF. The low-voltage d.c. obviously had a much bigger cap in the preamp chassis, maybe 4700uF or more.

[PRR]

Triode mode is lower noise, as HBP says.

Hanging V5 G2 on V7 cathode is like triode mode for DC, but Pentode mode for AC. Stable bias, high audio gain. Normally we can't do that very well: the large cap from a G2 resistor hanging on the pentode plate loads the audio. V7 is the same voltage but 10 times the current, little loading.

> consider measuring the step-up ratio of one of the input transformers.

The plan shows -60dBu in, -37dBu out, nominal 23dB gain, 1:14 ratio. Beyer did have a standard part of this ratio.

Assuming 150 ohm mike, the reflected impedance is 30K at grid.

To this they add 22K of dead resistance... a very odd thing. Conversely, the 22K reflects back as 110 ohms dead resistance in series with the mike, a heavy hit on Noise Figure.

It may damp the resonance between the high-Z winding and the high-C triode grid... but it seems like a wrong thing to do. A good clean triode off a studio mike can use 1:10 even 1:7 ratio iron and have good noise factor.

If it is the Beyer I know... it is very finely engineered but NOT big enough for modern studio levels. Anyway the overall gain suggests far-miked (TV) dynamics, not close hot condensers.

The final CF returns, not functionally through the shield, but through a link (happens to be shield) which will break if V5 is removed. It is perhaps intended to protect V7A grid-cathode diode.

The split/joint power is odd. However V7A V7B can draw high signal current, large signal on power nodes (as you say, 10u per node is not generous). ChA and ChB could be driving VERY different programs: sports and garden hour, or newsfeed plus talkback/cue to reporter. Or film-sound sync tone! Large isolation between outputs could be wanted. The input amps have, after the joint-filter, individual filters to kill crosstalk.

> pin 3 doesn't seem right

Born too late to remember when you never knew where an unbalanced XLR was hot. I forget when this was "settled". I know some makers kept doing it the other way.

> I assume carbon film given the era

I could be wrong, but I think Carbon Film was very new and only used for Microwave loads. The usual plate-load would be wire-wound. Ampex was special, and I'd suspect they implied metal film (though not the exotic metal films now marketed to audiophiles).

Yes, many odd details. This is a transition piece. Probably done in haste by a junior designer. 1950s Ampex did little that was surprising, but did it surprisingly well. 1960s Ampex did a lot of surprising stuff with those newfangled transistors. There was probably an influx of bright young men to support the change and expansion.

[RicharD]

Ampex had beautiful documentation.  I used their listed voltages for my DC analysis.  I hadn't gotten into the AC analysis and now HBP has done it all for me.

>The mic level is given as -60dB. Notice that after the input transformer, the level at the input grid is -37dB. You already got a 23dB boost just from the input transformer.

Cool ain't it.  Voltage gain, impedance matching, and differential to single ended (in this application) just from a couple of wads of wire wrapped around some nickle.  These input transformers are tiny, approximately a 1/2" cylinder maybe 5/8" tall.  As I think I mentioned in my 1st post, there is some debate as to the quality of these transformers.  Kind of a moot point because is one was to clone this unit, one would need to find a current production transformer.  We (Iso and I) are working on a tube mic preamp circuit, but it's nothing even close to this.  We're using transformers made by CineMag and wow they are sweet.

>I suppose you could add an output transformer to balance the signal if you'd like. That would be the way to do it...

Agreed.  Jensen makes an EXPESIVE off the shelf transformer for this purpose, Edcor makes an economical version, and CineMag now makes a moderately priced output transformer that they custom wound for us.    It would make sense that Ampex had a specific idea in mind.  It's about the same era as the 601 suitcase recorder.  Seems like a good mate for a portable recording system.  As for a hiZ out, it's actually quite low.   TubeCad spits a Zout of 436 ohms.  I 'd bet they put the signal output on pin 3 to put the output in phase when feeding a balanced input. Altec made a little 4 channel mixer too and it had an output transformer option.  The Altec 1567 is electronically vastly different from the Ampex MX-10 so there's hardly any point in comparing apples to oranges.

>I'm not 100%, but I think I've seen a whole regulated heater supply floated, and the artificial CT resistor used at the output cap to balance the circuit to ground.
I think I've seen something like that too.  I'm not certain I'm completely sold on series filaments.  Resistors are cheap and it only takes 2 additional ones to set a variable regulator.  Suppose 1 filament failed short, then the other tube's filament gets double what it's expecting.  2 die for the price of one.  Unlikely yes, but stranger things have happened.  I worked on a tape machine that used 4 14A8 (14V equivalent of the 6BM8) tubes with their filaments in series fed from a 56V primary tap (eek!).   Cheap... you betcha.  A pair of them were the tape bias oscillators.  A thought I had was if any one of these tubes failed, the whole unit pretty much quit working.  I'm leaning towards cheap being the reason.  In mass production, economics play a huge role.  In DIY, cheaper is not necessarily better, better is better.

[RicharD]

>The final CF returns, not functionally through the shield, but through a link (happens to be shield) which will break if V5 is removed. It is perhaps intended to protect V7A grid-cathode diode.

Simple, brilliant, & elegant.  I knew there had to be a reason, I was simply looking in the wrong direction. Idiot proofing 101.  I love it!

[HotBluePlates]

I'm not certain I'm completely sold on series filaments.

You regulate 1/2 the current when you output 12.6vdc instead of 6.3vdc. That means less heat in the regulator, and an easier design.

The big limiting factor will be making sure you don't exceed dissipation ratings for the IC if you use low voltage/high current.

[RicharD]

>I'm not 100%, but I think I've seen a whole regulated heater supply floated, and the artificial CT resistor used at the output cap to balance the circuit to ground.

I did a little reading, namely the cut sheet for the 12V regulator I have on hand, LT1084CT-12.
Cut sheet can be downloaded here
One of their example circuits shows the input bridge floating and the output of the regulator grounded thus creating a negative supply.  Seems like if it can sit on either side of the fence, it should be fine sitting right in the middle.  My thought was to ground pin 9 of the 12AU7 just like Ampex did but also ground the series connection of the 2 - 6267's.  This might be a REALLY bad idea whereas if one of the 6267's is pulled, the remaining one has a current path through 1/2 of the 12AU7's filament..... thinking as I type.... this would be bad.  Scrub that idea.  Either reference ground just through the 12AU7, or do a virtual center tap with a pair of 220 ohm resistors.  (220 ohm vs 100 to keep the current down, this is 12V not 6V)  That's a pretty simple experiment.  What's the worst that can happen?  ...... um...... I'm gonna do this experiment with some beater 6FQ7's from one of my proverbial buckets of used tubes vs. the sheik 6267's.  I won't cry if I lose a couple of them or a regulator .... but..... this doesn't really tell much about noise does it?  Shoot... I may just go ahead and build a complete single channel.  My board is stripped completely clean so I have no excuse... unless I have no PT.

>To this they add 22K of dead resistance... a very odd thing. Conversely, the 22K reflects back as 110 ohms dead resistance in series with the mike, a heavy hit on Noise Figure.

I didn't even notice that until you mentioned it.  I've been fedidelin with plate followers so much that this one slipped right on past me.  I can see no reason other than to kill high frequencies.  I'll be using a completely different input transformer so I'll omit that that resistor for the time being.

[RicharD]

Filament experiment complete.  It didn't make a difference where I referenced ground, virtual tap using 220 ohm resistors, pin 9 of the 12 AU7, or the series connection of the 6V tubes.  Better judgment told me to NOT pull a 6V tube when grounded at the series point.  I even tried mixing different brands (not types) of 6V tubes.  No troubles what so ever.  Seems like the virtual tap should be the way to go, but I guess I won't know that for certain until I have an actual audio circuit under test.  Theoretically since this is DC, noise shouldn't be a problem, but the proof is in the punch.

[PRR]

Why are you center-tapping DC??

That's one of the things on the original which made me think this was a hasty-re-work-up of some other bits and pieces.

Yes, tubes of the same nominal current should work in series. Series-string TVs demanded it, and the standard heater-stuff got used for almost all types even when not specifically rated for series use. Yes, TVs gave series a bad name, because they had a dozen very-different types (monster H-outs and teeny RFs) in string, all low-bid. Run 2 or 4 tubes in series you won't have a problem.

[HotBluePlates]

Ya know... if we read first and speculated second, we'd all look like electronical geniuses and stuff...

Ampex MX-10 / MX-35 Manual

[RicharD]

>Ya know... if we read first and speculated second, we'd all look like electronical geniuses and stuff...

Where's the sport in that?     Funny thing you guys nailed it.  Ampex documentation is excellent.  I especially like the sketches of people with Ampex equipment.

>Why are you center-tapping DC??

I guess it my day job standards influencing my thinking.  Funny thing is we discussed this very topic about 2 months ago except it was a 12.6VAC filament circuit.  Ultimately it was decided the filament circuit could be grounded anywhere, but the series point was probably easiest (IIRC).  Seriously, it's not all in 1 ear and out the other but I'll be the first to admit I suffer from SBD (slow brain disease).

[HotBluePlates]

The funny thing to me is that the checkout and adjustment section shows a HP 200AB (or CD) oscillator and an HP 400D AC meter, which was the setup I was thinking of when I mentioned them earlier (but just because I have that particular gear). Ampex spread the love a little in the maintenance section by replacing the meter with what looks like an RCA Senior/Master Voltohmyst.

It's also funny that the drawing shows the tech measuring the performance of the mixer with a cigarette in his hand. Times have changed...

[PRR]

> if we read first and speculated second, we'd all look like electronical geniuses

I think we hit all the points; just that Ampex said it more politely.

This thing cost a week's pay for a computer engineer of the time.

The noise figure is ample for any realistic use (and almost exactly what my favorite Bi-Amp mixer did 20 years later) but 5dB-6dB shy of theoretical (that 22K damper).

The big boast seems to be the high output levels, but who needs that much?

It filled a gap in the line-up of low-price (for Ampex) products. You sell suitcase recorders, you sell snazzy microphones; some guy is gonna wanna use more than two mikes. There were not a lot of alternatives in 1960; none to fit the Ampex suitcase. Gates probably had something, though probably not stereo. Langevin etc had better stuff, bigger, and 10++ times the cost.

The layout seems awkward.

While it may have been "thought about", IMHO it lacks elegance.

And yes, those mike trannies will mush modern studio levels. Trumpet through dynamic at 4 feet (as illustrated) won't stink, and any reasonable far-mike orchestra will be fine, but you don't track drums through large-condenser with this. Not just the trannies: with decent step-up in the iron the tube will be straining.

Many modern condensers will be fine if the inputs are replaced with 1:1 iron (600:600 or 10K:10K). With hot mikes, shielding is not critical.