5687 line stage

[kiira]

Hi... this seems like a neet way to get both gain and low impedance. If I have my maths right i(!!! :sad: ) it should give around 20dB gain and 110 ohms impedance. I've only seen 5687s used by sysmofit in the slowblow. I'd like to build this with a 6sn7 mic stage I guess. My maths skills suck so I hope this seems right... but I'm wondering about a couple of design questions I am really shaky on:

since the out put is low impedance would it drive phones, or +4 600 ohm linputs?

how would I determine what type of transformer to use with it for balanced output?

I am not sure how to figure out if a circuit meets the criteria for power I think is the problem.


larger image on my web page

http://www.2tough.com/~kiira/rec/5687linestage.jpg

thanks!

Kiira

 

[pstamler]

[quote author="kiira"]since the out put is low impedance would it drive phones, or +4 600 ohm linputs?[/quote]

Not well. The low-impedance output comes from the 100% feedback on the cathode follower, which reduces the output impedance by the same factor as the feedback. If the gain without feedback was 20dB, then you'd reduce the output Z by a similar factor, 10x. So far so good.

But think about what the gain without feedback would be if the tube was operating into 600 ohms. It would be quite low, and so the feedback factor would also be low, and so the impedance reduction (and distortion reduction) would be low. So you'd get an output that was higher impedance than you want, and higher distortion as well.

Norman Crowhurst has an excellent explication of this in his "High Fidelity Circuit Design". (I basically stole his.)

how would I determine what type of transformer to use with it for balanced output?

You'd really need a 1:1 transformer to maintain any gain from the circuit, and as described above you couldn't drive a 600-ohm load with it. But a 1:1 transformer would give you good balanced drive into 10k. a 10:1 transformer would give you good 600-ohm drive but no gain.

Peace,
Paul

 

[moamps]

[quote author="kiira"]
since the out put is low impedance would it drive phones, or +4 600 ohm linputs?[/quote]
The C2 is too small.

 

[CJ]

Plenty of power, that's for sure.

There is a 5687 phone amp at (gulp) Pat's site. :twisted: :twisted: :twisted: :twisted:

I am using it as a phone amp and it is very good for critical listening, but no headroom
and limited volume. There are alwas tradeoff's for fidelity.

 

[analag]

I have used that topology before, you don't need C1 both cathodes are in anti current phase so you pretty much have a constant current draw taking place right there, what you may need is a 0.1uF to prevent oscillation in the mhz frequency range. C2 should be larger. Try 10uF. Also the 5687 is one hot tube, use a good socket.

Analag

 

[kiira]

[quote author="analag"]I have used that topology before, you don't need C1 both cathodes are in anti current phase so you pretty much have a constant current draw taking place right there, what you may need is a 0.1uF to prevent oscillation in the mhz frequency range. C2 should be larger. Try 10uF. Also the 5687 is one hot tube, use a good socket.

Analag[/quote]

Hi analag, just curious what did you use it for?

ok I will make that cap a 10uF .... but what will that accomplish?

thanks everyone for the help

Kiira

 

[kiira]

[quote author="pstamler"]

Not well. The low-impedance output comes from the 100% feedback on the cathode follower, which reduces the output impedance by the same factor as the feedback. If the gain without feedback was 20dB, then you'd reduce the output Z by a similar factor, 10x. So far so good.

But think about what the gain without feedback would be if the tube was operating into 600 ohms. It would be quite low, and so the feedback factor would also be low, and so the impedance reduction (and distortion reduction) would be low. So you'd get an output that was higher impedance than you want, and higher distortion as well.[/quote]

I see. So the unsuitability for driving a load is a function of the design itself and cannot be gotten around.

Norman Crowhurst has an excellent explication of this in his "High Fidelity Circuit Design". (I basically stole his.)

Is that a book Paul? I have read all the stuff by NC that audioxpress has on their site - he explains things pretty clearly.

thanks!

Kiira

 

[PRR]

I do not understand....

Why you link to a large image and then put a 800-wide image IN the post?
I can't read the words without scrolling side to side; and I won't.

Why the voltage-amp stage is flowing as much current as the power-output stage?
The later drives heavy loads, the former just drives a teeny grid.
OK, as analag says this cancels supply current variation;
it does not improve PSRR which is only 5dB-10dB.

Why 8.25K?

What the "152" cap in the bottom does? No unit... is that Farads????
If you leave it out, you have positive feedback... Why?
Maybe not enough to notice: you don't have enough forward gain,
especially with that low-low plate resistor on V1.

Why it says "135K" at IN?


> would it drive....

Without knowing much about the tubes,
just pretending you got them biased into the active zone:
the output tube must be flowing (or trying to flow) (330/2)/8K= 20mA.
For good low distortion, the change in current with signal
should not be 2:1 away from the idle current.
That says, for 20mA standing current, swing down to 10mA and up to 30mA,
or 10mA peak in the load.
The 8K (and the tube internal resistance) is part of the load:
10mA peak in the 8K is 80V peak, so we can't get 80V swings
while driving any additional load.
Assume 40V peak swings with half of 10mA or 5mA in the load.
The load can be 40V/5mA= 8K, the RMS voltage can be 40/1.414= 28V.
Maximum output power is 100mW or +20dBm,
about right (bit not generous) for nominal "+4dBm" audio interfaces.
That's assuming you have an 8K:600 transformer.
Of course now the overall gain is 14dB.

To get this 100mW or 0.1 Watt output,
the plate power is 2*0.020A*330V= 13.2 Watts.
Plate circuit efficiency is 0.75%, compared to
50% for a perfect transformer-coupled stage,
8% for a perfect resistor-coupled stage, or
2% for a reasonably practical resistor-coupled stage.
Or to look at it the other way:
your power supply is 3, 10, or 30 times bigger than it could be
(though the best efficiency requires fancy iron).

With a 10K:10K transformer,
it would drive modern 10K loads far past the +20dBu level.
Assuming 10K loading, the output cap reactance has to be
less than 10K at (say) 20Hz.
Reactance chart leads you to 1uFd, not 0.47uFd.
In fact the transformer impedance rises with frequency
while the cap's impedance falls.
Because the transformer is also about 10K impedance at 20Hz,
it together with a 1uFd capacitor will resonate.
It is a series tank so its total LC impedance
wants to fall to zero at resonance.
This will upset the tube and kink the frequency response.
Probably a huge peak at the bottom of the audio band,
but only for very small signals;
larger signals will just distort bad.

 

[alk509]

[quote author="kiira"]I will make that cap a 10uF .... but what will that accomplish?[/quote]

Good bass response when driving tougher loads. The output cap forms a HPF with the load impedance it's connected to... So if you plug your circuit as-is into a 1K input, the filter's f-3dB = ~340Hz - not good. With a 10uF cap at the output, f-3dB = ~16Hz - much better.

At this point, of course, someone who really knows what they're talking about will show up and say I'm wrong...

Peace,
Al.

 

[pstamler]

[quote author="kiira"][quote author="pstamler"]

Not well. The low-impedance output comes from the 100% feedback on the cathode follower, which reduces the output impedance by the same factor as the feedback. If the gain without feedback was 20dB, then you'd reduce the output Z by a similar factor, 10x. So far so good.

But think about what the gain without feedback would be if the tube was operating into 600 ohms. It would be quite low, and so the feedback factor would also be low, and so the impedance reduction (and distortion reduction) would be low. So you'd get an output that was higher impedance than you want, and higher distortion as well.[/quote]

I see. So the unsuitability for driving a load is a function of the design itself and cannot be gotten around. [/quote]

Well, yes and no. As PRR points out, you could use an 8k:600R transformer and drive a load nicely, but then your gain is less, about 11dB less if I calculate right. So your overall gain would be about 9dB if the original gain was 20dB. You could get around that by using something like a 12AX7 or, better, a 6SL7 as the first stage, which would bring your gain up to something reasonable.

However, you'd run into clipping problems. Assuming you want to put out +20dBu into 600 ohms; given the tranny, that means your cathode follower needs to be putting out about +31dBu. In my experience it'd be well into clipping by then.

Here's a possibility: see if you can find a 4k:600R tranny. That has 3dB less loss, so now you only have to put about +28dBu into it. Difficult but not impossible; to drive the lower impedance load, you'd put the two halves of the 5687 in parallel, using a second tube (again, I'd suggest a 6SL7) for the voltage amplifier.

Juggling like this is why push-pull outputs (as used in Ampex recorder outputs) and totem-pole circuits got popular.

Norman Crowhurst has an excellent explication of this in his "High Fidelity Circuit Design". (I basically stole his.)

Is that a book Paul? I have read all the stuff by NC that audioxpress has on their site - he explains things pretty clearly.

Yes; looking at the copy on my shelf I discover that in fact Crowhurst co-authored it with George F. Cooper; it was published by Gernsback in 1956, and was (at least in part) composed of reissues of articles they wrote for Radio-Electronics, which later became Electronics World, merged with Popular Electronics and more-or-less disappeared. Look on www.abebooks.com for a used copy -- and good luck. Or use interlibrary loan -- I'm sure there's a copy out there in a university library somewhere.

Peace,
Paul

 

[analag]

[quote author="kiira"]Hi analag, just curious what did you use it for?
ok I will make that cap a 10uF .... but what will that accomplish?
Kiira[/quote]

I used it as a line stage pre, seems to have a very extended top end, average PSRR. I always use a large valued cap on the output of a CF it provides a more consistent bass response with a wider range of loads. Use polypropylene not electrolytic, Solen is my choice for big cheap polyprops.

 

[alk509]

[quote author="analag"]Solen is my choice for big cheap polyprops.[/quote]

Hear, hear!

Peace,
Al.

 

[CJ]

You could also use a 6900 (Bendix version) if you can find one.
Gm is 11,500, compared to 8,500 for the 5687. Or the GE 7044, which comes in at 12,000, although not as rugged.

Or the E182CC, European version with a Gm of 15,500 might be cool if you come across one. Brought into US as Amperex 7119.

Warning! Do not plug 5687 into 12AU7, 12AX7a or similar socket. Major damage if you blow it.

"Large cathodes and careful manufacture make these tubes suitable for audio use, especially as cathode followers or for fixed line-level stages. The 7892 was a remarkable device for it's small size; it's rated to produce 5 amperes in short duty cycle pulses. All the 5687 types are similar in general ratings to the 6CG7 or 6SN7GT, while having the high transconductance and high perveance that makes them technically excellant classics. If you have a stock of them, use them in your homebrew equipment. Do act quickly, though, and stash away a lifetime supply, these types are disappearing rapidly. The only one still being used by the US goverment is the 6900, and even that is declining in numbers as old avionics are scraped. MU still manufactures it occasionally, making it one of the very few recieving tubes still in production in the USA."

Eric Barbour

 

[PRR]

> if I calculate right.... about 9dB if the original gain was 20dB.

I need to oil my abacus. Your numbers are more-right than mine.