M10a reverb question

[VMS]

Hi guys!

I'm wondering about this M10a reverb circuit.

What is the gain difference on dry signal between Vin and Vout?

I know 15k and 2k7 makes a voltage divider but is there gain when signal goes in the cathode and comes out the plate?


thanks in advance.

[sluckey]

I never measured the input or output signal levels so I can't say what the overall gain is. However, you do get gain when you put a signal into the cathode and take it out on the plate. The gain will be the same as if you had put the signal in on the grid, but there is no phase inversion.

[HotBluePlates]

What is the gain difference on dry signal between Vin and Vout? ...

I think I see a Dry gain of about 0.6 (a loss of about half), based on a half-assed analysis.  Maybe PRR can tell me if I'm on the right track:

There's a big signal loss on the first 12AU7 stage due to taking a cathode output across such a small load.  I didn't try to formally analyze the stage, but since plate current = cathode current, the cathode output is likely only 270Ω/5kΩ = ~1/18th the size of the plate output.

As you note, the 15kΩ forms a voltage divider with the 2.7kΩ cathode resistor at the 12AX7 mix stage, which is another 2.7kΩ/(15kΩ+2.7kΩ) = ~1/7th loss.

I can't trust the 12AX7 internal plate resistance is the data sheet value of 62.5kΩ, but rolling with that number the gain of that stage is 100 * [270kΩ*(270kΩ+62.5kΩ)] = ~81.

1/18 * 1/7 * 81 = ~0.6

Of course the Reverb path has a hefty signal loss through the reverb tank, so it seems reasonable it would be comparable to the Dry path signal, even with the extra 12AX7 gain boost for the Reverb signal.

[PRR]

> Maybe PRR can tell me

Better than that.... my father is looking over my shoulder, astonished to see grids and cathodes on a web page. Earlier he was telling of his modified Cathedral radio with push-pull type 45 power and a type 80 rectifier.

Yeah, yeah, you can calculate with tolerable accuracy. But if it is in your hand, you can measure it. If this is hypothetical, you might try assuming the Original Designer did his job, and if this is inserted in an existing signal path then Unity Gain would be a very good bet. Finally, if you don't like the gain as-is, you can diddle the 15K to 5K or 50K and get some other gain. You don't always need an exact answer to use a plan.

I'll be a bit shy for a few days while my folks are in town.

[PRR]

Gain of bare cathode follower is unity.

But all the resistance in the plate circuit, divided by Mu, appears as cathode resistance.

Taking one 12AU7 at high current as Rp=6K, we have 3K, plus the 5K, is 8K. Divided by Mu of 20, is 400r.

400r to the 270r resistor is gain of 0.4.

At the grounded grid end, if we ignore the 2.7K, gain is effective resistance on plate node to the 15K source. 12AX7 at low plate current is say 100K Rp. With 270K plate feed, say 75K here.

So GG stage gain would be 75K/15K= 5.

But the 2.7K diverts much of the signal from the 15K. I'm having trouble proving this, figuring what factors cancel out. But assume the 270K is divided by Mu, gives 2.7K, so the current in 15K is split about in half.

0.4 * 0.5 * 5 = unity.

That is assuming NO load at Vout. Impedance here is ~~75K, so even a light load will be some drop. A complete study would need this data.

[sluckey]

There is plenty of gain in other circuits in this amp. It would make sense for the reverb circuit to have an overall gain of unity. I can measure the real world gain but it would be a lot of trouble. Just how important is it to know?

[VMS]

Thank you guys. Learned something new about cathode followers and was it grounded grid gain stage.

I can measure the real world gain but it would be a lot of trouble. Just how important is it to know?

Not that important. I was just thinking maybe using this reverb block on some other amp circuit and if there was a significant gain loss that would have to be compensated.

What about this block from M8, is this closer to unity gain than the M10a block?

I could use 12dw7 tube on this one.

[HotBluePlates]

What about this block from M8, is this closer to unity gain than the M10a block?

Gain of the 7025 stage is about 88.

The Dry signal sees a 470kΩ resistor leading into a combined 27kΩ (pot & 22kΩ resistor).  With the Mix set for max-Dry, the reduction is 27k/(27k+470k) = 1/18.  So net gain of 88/18 = ~5 for the dry path.

[VMS]

Thank you sir.

I found another M8 schematic from this site:

https://irationaudio.com/2015/10/31/magnatone-m8/

This is similar but the reverb is cathode follower driven. Do you guys think this one is a revision version to make the reverb stronger?

Would these circuits use the same reverb tank?

[HotBluePlates]

This is similar but the reverb is cathode follower driven. Do you guys think this one is a revision version to make the reverb stronger?

Would these circuits use the same reverb tank?

The tanks are likely the same, because they are cap-coupled and likely provide a relatively-high load impedance to the driving tube, as compared to the relatively-low impedance of the transformer-coupled reverb tanks common in Fender amps.

Was it intended to make the reverb stronger?  Hard to say, as the important element for driving a reverb tank is the current-per-turn in the tank's input transducer.  I'm a bit unsure if the idea behind the change was an attempt at more output current, a bit wider bandwidth, or what.

As suggested earlier, the best approach would be to breadboard the different circuits and measure how they perform.

[Willabe]

They might have wanted to save the $ on the verb OT and chassis space?

[2deaf]

The M8 schematic shows 5K across the reverb output coil.  Nowadays, it is hard to find a tank with anything but a 0.358H output coil.  5K ohms and .358H gives a cut-off frequency of 2.2KHz, so the highs will be sheared off of your reverb sound and you probably won't like it.

The M10A has 47K for the output load with a cut-off frequency around 21KHz.  It will still cut some highs in our range of interest, but it will be minimal.  If you parallel the 47K with a .0022uf capacitor in a effort to get a brighter reverb, you will get the opposite effect because you will have fouled-up the resonant frequency phenomenon upon which the brightness effect depends.  You would need to raise the output load to 220K in order to get the resonant frequency brightness effect, which looks like no problem for that circuit.

[2deaf]

The Dry signal sees a 470kΩ resistor leading into a combined 27kΩ (pot & 22kΩ resistor).  With the Mix set for max-Dry, the reduction is 27k/(27k+470k) = 1/18.  So net gain of 88/18 = ~5 for the dry path.

Look at that schematic again and perhaps lower your net gain result by an order of magnitude or so. 

[HotBluePlates]

The Dry signal sees a 470kΩ resistor leading into a combined 27kΩ (pot & 22kΩ resistor).  With the Mix set for max-Dry, the reduction is 27k/(27k+470k) = 1/18.  So net gain of 88/18 = ~5 for the dry path.

Look at that schematic again and perhaps lower your net gain result by an order of magnitude or so.

Well, the error was going with an rp of 62.5kΩ from the 12AX7 data sheet.  But that's because I didn't want to draw on a sheet to get an accurate rp, which turned out to be ~139kΩ.  And Mu is down to 90, while the effective load is only 320kΩ once you factor in the 1MΩ which follows.  All told, the math says a gain of 90 * [320kΩ/(320kΩ+139kΩ)] = ~63, while following the a.c. loadline looks more like 60.

If you were talking about using 27kΩ in the formula rather than the pot's 5kΩ, make sure you didn't overlook that 22kΩ to ground...  The Dry never drops to zero-level, even though the pot is ganged to reduce dry signal when the Reverb is turned up.

[2deaf]

If you were talking about using 27kΩ in the formula rather than the pot's 5kΩ, make sure you didn't overlook that 22kΩ to ground...  The Dry never drops to zero-level, even though the pot is ganged to reduce dry signal when the Reverb is turned up.

I meant the two 3.3K resistors in parallel with the 5K + 22K  making the divider 470K to 5.3K.  The 3.3K to 3.3K then cuts the signal in half at the grid.  This is at max dry condition.

[HotBluePlates]

Yes, additional loss through the mixing resistors.  Thanks!

[2deaf]

I have one more point to make before this thread disappears into oblivion.

The M10A driver can deliver 26V_pp at the point of clip into a 750 ohm load at 1KHz with an E_bb of 300V.  Pretty impressive.  HOWEVER, it idles at 20.0 ma DC.  What a pig.

There is always a trade-off between current consumed and reverb driven, but 20ma is just too much.

[VMS]

I have one more option to consider. The reverb from magnatone 431.

But this one has 1k resistor across the reverb output and reverb control is little odd, so I'm thinking maybe taking the recovery part from the m10a circuit.   

[PRR]

> 20ma is just too much.

It depends??

On a cold winter day, I'd appreciate the extra 6 Watts of heat. (Conversely in summer.)

If this goes in a 4x6L6 Super Twin eating 450mA at full roar, another 20mA looms small.

If this goes in a Champ.... well, an odd thing, we build Champs with DeLuxe PTs capable of feeding another 6V6 at ~40mA. So it doesn't really impact the power supply budget. (Some, because this 20mA needs more filtering than a 6V6's plate.)

Figuring the Electric Bill, the 6 Watts is another tenth-cent per hour. Or a buck every 1,000 hours. If you don't expect that much "joy value" from a reverb, you probably would not put it in.

There is the alternative. Use a transformer. While audio transfromers have become Specialty, we do have the perfect part for *this* application, the Fender 12AT7-tank RT. Like 1/4 the DC power for a couple dozen bucks?

[2deaf]

Why hold back?  Switch out the Princeton PT in your Champ for a DeLuxe PT then drive the reverb with a 6V6 in pentode mode.

[2deaf]

I have one more option to consider. The reverb from magnatone 431.
But this one has 1k resistor across the reverb output and reverb control is little odd, so I'm thinking maybe taking the recovery part from the m10a circuit.

All you need to do is replace the 1K resistor across the reverb output with a 220K resistor and wire the Reverb pot. like the M10A.  The 12DW7 mixer stage is the 12AU7 half of that tube, so the unused half of the 12AU7 that you are using to drive the reverb works here.

The DC current consumed by the whole 12AU7 is 9mA.  Of course this driver won't deliver as much current to the tank as the M10A, but it will probably deliver at least twice the published nominal current for an 8E tank.

The 0.22uf capacitor between the 12AU7 plate and the input coil is way too small for an 8E tank.  The resonant frequency for that cap. and an 8E tank is 951Hz at which point the impedance seen by the tube is extremely small.  The tube will overload the input coil around 951Hz making an ugly signal and a bad sound.  The 1uf used by the M10A is also too small with a resonant frequency of 446Hz.  10uf would probably be okay especially if you cut the lows before you get there.  22uf would be better.       

[VMS]

Why 8E tank? I thought these circuits would use 4/8/9FB... type tank.

10uf would probably be okay especially if you cut the lows before you get there.  22uf would be better.   
   

I think one problem with this circuit is that you can't cut lows without affecting the dry signal.

[2deaf]

Why 8E tank? I thought these circuits would use 4/8/9FB... type tank.

Because at some point in the past I analyzed the M10A driver using an 8E tank so that I already had the numbers I posted in this thread.  I think that a type "F" tank is better suited for this driver and I probably used a type "E" tank because I didn't have an "F" at the time.

There is still a problem with the resonant frequency with a 4F tank, it just shifts to lower frequencies.  With a 0.22uf it's 700Hz, with a 1uf it's 328Hz and with a 10uf it's 104Hz.     

[terminalgs]

The photographed schematic is earlier than the one with a white background (see sequentially issued schematic PN 66-3065 vs. 66-3095).  The later one would have been what the engineers thought was superior (and it actually is very similar to the 460 that pre-dated the M8.  I suspect the 66-3065 had some issues so they reverted the best they could to the 460 style.   the 460 used a 6DR7 instead of a 12AU7.  The 6DR7's were more expensive than 12AU7's but that triode on the 1,2,3 side of the 6DR7 was better suited as a driver for reverb tank than 1/2 a 12AU7.  If you compare the 460 and the M8 schematics you can see they were playing with the load of the 12AU7 driver side.


reasoning: saving money. Estey was continually dealing with cash flow problems and filed bankruptcy a couple times in this era.  In you look at the tubes used in the '61-62 400 series amps, the engineers were having a field day with the new fangled tubes and tube manufacturer datasheets, 6DR7's, 6CG7s, 6EU7's, 12BH7's 12DW7's, etc all in the same amp.  with the suitecase line  it was all rationalized down to 12AX7s,12AU7s, and 12DW7s for all the amps.  If budget wasn't an such an issue for the M8, they would have driven the reverb tank with a paralleled 12AU7 like they did on the bigger, more expensive M10 and M13 amps.  The suitecase amps were really a marvel of component rationalization and cost saving measures that didn't lower the engineering quality of the amplifier.


I haven't been able to verify the reverb tank used in M8's, but I think it was the 25-0002 (Estey PN), which estey used in most of their reverb amps including the 460, M10A, M13, 450A, and 440.  the 25-0002 was around Zin~=1.2K, Zout~=2K. the other tank they used was a 25-0005 with Zin~=8ohm and that was driven by a transformer (some M15s and MP-1/3/5)