Raytheon 5784WB as sub for AC701?

[PRR]

> the AC701K is highly photosensitive

It (and most tubes) are slightly photosensitive; worse in this application because (as you say) grid current matters.

The photo-problem was only noticed when photographers used flash-bulbs close to the mike. If you build a very good mike, and put it front of Famous People, that happens. But most of us are not worth photographing.

> the 5784WB has a grid curent of 300,000 pA

That's Maximum, not typical. 90+% of 5784 will have much-much lower Ig; at $5/each, we can afford to sort-out the worst leakers, and hand-trim the remaining bias-error.

Will it be as-excellent-as AC701? No. Will it work? Yes, and perhaps quite well. It gives a $50-$500 mike instead of a $500-$5,000 mike. For many of us, that's just our level.

 

[riggler]

Gus,

Thanks for the nod on the C12 PSU. I take it you brought it up for the heater supply design? Nifty use of the zener there, and it looks to be well-regulated.

Tubologic,

You make some very good points (and give us some very valuable information in regards to other tubes and the shortcomings of non-AC701K subminis). But let me re-iterate that we are not doing a "drop-in" replacement here. We are, in part, modifying the architecture. After heater voltage and grid voltage mods, we'll be running the 5703WB in a fairly linear curve at an acceptable plate voltage. PRR addresses loading...

5703 sheet says design-max grid resistor is 1.2 Megs. The microphone circuit has 150 Megs, and for full bass and low bass noise you need something this large. '701 was designed for very-very-very low grid current. Other tubes have "low" grid current, but not tested past the point that would make trouble in a typical 1Meg resistor. Problem? Probably not. The 1 Meg spec is worst-case: bad day on the sealex machine, and a socket which pushes other parameters to the limit. If you aim for the edge of destruction, a large grid resistor and a poor-vacuum tube may put you past the edge. But the low supply voltage and large 100K resistor mean there is no great power dissipated or available. So most tubes will bias-up fine even with "over spec" grid resistor; those that don't will just bias wrong, won't smoke or melt.

So, can we say this is every bit as good as an implementation as the AC701 in this circuit? Of course not. And, based on your advice, maybe there is a better "sub" than the 5703. I'm going to check out some of the numbers you listed. Like Magneto said, there have been good implementations using these subminis out there. So, I'll give it a go and see what happens.

 

[riggler]

Please check my math here...

If the input heater is 6.3 volts, assuming R4 remains 200K, I compute R5 needs to be 850k to arrive at 5.1V at the grid node.

This would give us -1.2V bias. What the heck is R1 for?

All good???

 

[rodabod]

If you really feel that these JAN sub-minis are as crappy as all that, perhaps you should tell all the manufacturers (Neumann included) that are making good use of them in their current products.

I do believe Neumann did have issues with using subminis. Was it the M147? Anyway, Oliver Archut documented this.

I'm with PRR on this one though; these submini valves can work perfectly well in my opinion with respect to noise and overall audio quality. There are some nice Russian options out there too don't forget.

 

[PRR]

> 6.3 volts, ...R4 remains 200K, R5 needs to be 850k

Looks good as-stated.

Food to chew: As mentioned, grid current IS an issue, and is somewhat affected by cathode temperature. And you are running a 15mA tube at 0.75mA, like 1/20th of the maximum cathode emission. Under-heating is an option. 6.0V will work as good as 6.3V, no danger to the cathode. Dynaco ran 12AX7 at similar currents as cool as 5.5V heater.

Of course as you dink with heater voltage, in THIS plan, you indirectly change grid bias.

So I might plan for 6.0V, or maybe 5.6V (3-pin 5V regulator with diode in ground leg).

And since the "-1.2V" is a "typical", and you have wildly violated the 1.2Meg max grid-R spec, you will probably have to trim it in-circuit.

> What the heck is R1 for?

The capsule is 75pFd. It is equivalent to a perfect generator plus 75pFd series capacitor. You may analyze it as a "coupling capacitor". If choice of cap were unrestricted, and grid resistor is the safe 1Meg, I would toss a 0.01uFd (10,000pFd) cap for 17hz response. But you are "forced" to work with the 75pFd. For same bass limit, you need a grid resistor 10000/75= 133 times bigger, 133Meg. 150Meg is a rounder number, and gives -3dB at 15Hz and -1dB at 30Hz.

Ah, R4 R5 add to this value, but the "error" is trivial.

In this plan, we also want a 4V:1.6V (or 6.3V/1.2V) voltage-division.

For minimum parts-count, we could use two resistors instead of three. The ratio must be 6V.3:1.2V, and the parallel equivalent must be near 150Meg. I get values like R1=short, R4=187Meg and R5=793Meg.

In many jobs, a resistor saved is 12 cents earned. However many-Meg low-low leakage resistors cost more. They are not readily available in odd values. And the DC current through the huge-value divider can incite excess noise. Hence one special 150Meg and two cheap hunnred-K resistors.

 

[strangeandbouncy]

Just being curious,



  what effect would a flash-bulb have had to a mic with non-protected tube? is it expressed as noise or level change, or what?



      Kindest regards,


  ANdyP

 

[riggler]

And since the "-1.2V" is a "typical", and you have wildly violated the 1.2Meg max grid-R spec, you will probably have to trim it in-circuit.

....ehh, another newbie question on this point... So, I use a ~150k trimpot in place of R1 to achieve this? Or is it better to physically change resistors? Either way, I'd be adjusting the value of R1 while using a DC volt meter where the arrows indicate on the schemo, trying to attain -1.2V, right? (I know it's blatantly obvious but I gotta ask anyway...)

The capsule is 75pFd. It is equivalent to a perfect generator plus 75pFd series capacitor. You may analyze it as a "coupling capacitor". If choice of cap were unrestricted, and grid resistor is the safe 1Meg, I would toss a 0.01uFd (10,000pFd) cap for 17hz response. But you are "forced" to work with the 75pFd. For same bass limit, you need a grid resistor 10000/75= 133 times bigger, 133Meg. 150Meg is a rounder number, and gives -3dB at 15Hz and -1dB at 30Hz.

You used this formula to determine "I would toss a 0.01uFd (10,000pFd) cap for 17hz response."?:

I get 15.9Hz! 

Anyway, I GET IT!!

For minimum parts-count, we could use two resistors instead of three. The ratio must be 6V.3:1.2V, and the parallel equivalent must be near 150Meg. I get values like R1=short, R4=187Meg and R5=793Meg.

In many jobs, a resistor saved is 12 cents earned. However many-Meg low-low leakage resistors cost more. They are not readily available in odd values. And the DC current through the huge-value divider can incite excess noise. Hence one special 150Meg and two cheap hunnred-K resistors.

Read you loud and clear, thank you!

 

[PRR]

> I use a ~150k trimpot in place of R1

R4. Or R5. R1 is a magic high-high value part which you do not want to disturb, not even touch. (Finger-damp will change its value.)

It's a two-step process. You have to take the guts out to probe voltage and change resistors or lash-on a trimpot. But you eventually have to put it together, and I assume it is too tight to cram-in a haywired pot. So find the right value with whatever temporary method, then do it neat enuff to close.

> You used this formula

No, I don't keep a GIF image of a formula in my head. If I did, it would be a Reactance Chart.

I can remember a few things:
10uFd+1K = 20Hz
0.01uFd+1Meg = 20Hz
0.01uFd+1K = almost 20KHz

Transpositions of these will estimate most any audio R-C problem.

> I get 15.9Hz!

20Hz, 17, 16, 15.915494309189533576888376337251Hz.... Not only am I a 20% guy, the "answer" is for a semi-arbitrary 3.0102999566398119521373889472449dB corner. Fussy-folks may ask for the 1dB point. Others may accept -6dB. That's an octave either way of 15.915...Hz. Remembering "20" is easier, and usually conservative (for bass; I need to remember that 0.01u+1K is significantly under 20KHz).

 

[riggler]

Sorry PRR,

I meant to say R5! If I can physically FIT a trimpot in there, is there any reason not to? Eeh, I can see it already, gets a little oxidized, etc, changes value, destroys tube...

Better just solder in the final resistor...

THANKS.

 

[Rossi]

You don't want any significant amount of DC running through a pot. What may work is replacing R4 and R5 with one fixed resistor  and connect a large value trim pot (1 meg or more) in parallel to that fixed resistor. Then connect R1 to the wiper of that pot.
That's how it's done in the Schoeps schematic, but that's a FET circuit, of course, and due to the very large gate resistor (1 Gig), there's very, very little DC through the wiper.

 

[Rossi]

[...] This will make the difference between a average microphone and an exceptional one,like the old NEUMANN's.
To summarize: The AC701K was an outstanding but peculiar tube,don't expect to replace it with a cheap hearing aid or shell proximity fuse subminiature tube. Both are small and made of glass (and metal),but that's all they have in common.

Thank you tubelogic; that sure was a very informative posting! But then again, not all old Neumanns used an AC701. In fact, even while the AC701 was around, Neumann and others used other, more common tubes in most microphones not intended for the (German) broadcast market. The U67 used the EF86. AKG/Telefunken used the 6072 in the American version of the ELA M251 and in the C12. Gefell used the EC92. Today's tube mics using subminis are among the lowest noise tube mics ever produced, so they can't be totally bad.
That's not to say all those tubes are as good for microphone use as the AC701, but there's life before and after that tube. Part of the reason why the AC701 looks so superior on paper may be the fact that it was about the only tube explicitly produced for microphones, so it was manufacturer specified and tested just for that. With other tubes the datasheets were formulated for more basic tasks - radios etc don't need grid resistors higher than 1 meg, so why test or specify for more? That would only increase the number of rejects. So for microphone use you have to do your own testing and reject tubes that do not meet the criteria. In fact, that's what Neumann and others did with all those more common tubes. Part of the reason why the U67 was discontinued was the fact that the reject rate for EF86 tubes went sky high. For guys like us, the situation isn't quite so bad. We don't want to produce hundreds or thousands of mics. And if a tube is not good enough for a microphone, we may still be able to use it in another project.

 

[tk@halmi]

The 5784WB makes a cheap and pleasing sub in the Royer 1 type circuit. A few db less gain using the stock circuit.

 

[rmburrow]

The 5703 was used in the Capps CM2250 series.  The 7586 nuvistor was used in the Neumann U64 and AKG C12 US versions.  The EF86 was used in the Neumann U67.  The 5703 may be the closest "substitute" to a AC701 with the exception of the heater voltage and the separate grid connection.
Haven't tried the 5784 in triode.  The 5840 in triode was used in the Altec "lipstick" condenser mics.

Done carefully, a 7586 can replace a dead AC701.  By dead, I mean open heater on the AC701.  Do not "condemn" a AC701 that has heater continuity before checking the associated components, preferably by substitution.  Power supply:  Stabilytes (in the NKM and NSM supplies) can be replaced with NiCad batteries.  Use 3 for a AC701 and 4 for a 7586.  If you replace stabilytes, RECHECK the heater voltage carefully.  Use a calibrated digital voltmeter with a load on the heater supply.  Set the supply to 3.9 volts for a AC701 and 6.0 volts for a 7586 at the tube.

If the AC701 is indeed bad, the 7586 mod must be done in a reversible manner.  Orient the tube socket with the grid toward the capsule; make the grid connection through a short (app. 1 cm) No. 20 bare wire.  Bundle the heater, cathode and plate connections away from the high impedance grid and reuse the terminations used by the AC701.  Change the plate resistor to 100 k; make the cathode resistance a divider of 560 ohm and 2200 ohm resistors in series. Connect the "hot" end of the 20 k/30 k resistor grid bias divider (in a Neumann) to the junction of those resistors.  The heater voltage for the 7586 is 6 v dc. Ground the shell of the 7586 by coiling 3 to 4 turns of No. 18 bare wire around it and connect to suitable ground.  NOTE this mod will NOT work on a C60 or other mic where the capsule polarizing voltage is derived from the IR drop from the cathode to ground.

Last but not least, write (preferably) or email JJ Electronics and Eurotubes and request they research manufacturing the AC701 at their facilities.  Their EF806's are quiet.  Have two of them for use in U67's.  If you order tubes from Eurotubes, they want to know the equipment they will be used in.  Good first start.  Should JJ Electronic build good quiet AC701's from the Telefunken design and sell them for $100 to $200 a pop, that is a lot better than present prices.  In 1969, AC701's could be bought for around $10 to $15 each.

 

[ricardo]

FWIW, Bernhard Weingartner, who designed the C12 for AKG, hated AC701 and much preferred the 6072 used in practically all export C12s.

I have this from Clem Beaumont, my mike mentor at Calrec, who was a personal friend of Bernhard.  It was Bernhard's encouragement of Clem's DIY efforts that led to Calrec.

 

[Bill Wilson]

Hi All: I have an AKG C60 mike system; I had rebuilt the power supply (which cured the hum from the original filter capacitors age) & sold it. The buyer complained of a very low level hum; took it to a (supposed mike guru for AKG in L.A.), he could not find the problem.
Got the mike back & solved the problem, (it was the LM317 regulator ckt. that replaced the original filament voltage regulator). Got rid of the noise w additional filtering & ready to send it back to the customer; as I was checking the ac connector my meter probe slipped, shorted out the AC & blew the fuse. When I repowered the mike, the AC701 was dead, open filament!

Found this thread & contacted Magneto Sound; he was a great help to me. I already knew the cathode follower should not affect the sound of the capsule. As a precaution I checked with Richard Land in Nashville, (he was the national microphone repair manager for AKG, when they were located in Nashville & now repairs AKG mikes as Land Audio Services). Richard agreed with my opinion, that the cathode follower should not change the sound of the original C 26 capsule. I did have to change the cathode resistor from 1.5k to 2.5k
& alter the dropping resistors in the plate voltage filter chain in order to arrive at the final result.

Bottom line: the C-60 now uses a 5784 tube. These are the measurements: Filament 5.6vdc, plate 107vdc, capsule polarization 55vdc, cathode bias 1.7vdc. I adjusted the filament for lower voltage in order to minimize tube noise. The mike sounds the same as with the AC701. I also have the article by Dave Royer on modding a Marshall 603 mike using a cathode follower output. He gives a detailed description of cathode followers used in mikes, confirming the above.

 

[Bill Wilson]

One other detail: I could not wire the 5784 as a triode, it would not work. 5784 is wired: grid #2 is tied to the plate & grid #3 is tied to the cathode. It is wired as a pentode.

 

[PRR]

What is "would not work"?

> grid #2 is tied to the plate & grid #3 is tied to the cathode. It is wired as a pentode.

If grid 2 is wired to the plate, then it works as Triode for practical purpose. Things are very different when the accelerating electrode (G2) is held *fixed* and the collecting electrode (plate) swings around.