ECC82 based preamp

[arnoldc]

Hi Giaime, how do you change the current setting of your CCS?

 

[planet10]

This basic topology is best exemplified today as the Bottlehead Forplay ... common cathode direct voupled to cathode follower. It also uses 12AU7 (as already pointed out, has more potential with 6SN7 -- Frank posted a couple maps for that one). A common improvement made on them was a CCS anode load on voltahe amp & a CCS on the cathode of the follower.

dave

 

[Giaime]

Hi Giaime, how do you change the current setting of your CCS?

First I would thank Greg the Geek for having it explained to me. Let's call the upper resistor Rb, and the lower one Ra.

Quoting Greg:

Hi Giaime,

1) if I understand right, that circuit doesn't need a separate supply (that's a good thing for me)?

Correct.

As long as the voltage across the entire device is greater than about 1.7V, it will do its job.

2) i have to sink 8-10mA, would 2 BC109 work? I also have 2N2222A, BC161, 2N3820, 2N2907...

At 8-10mA, almost anything will work, as long as Vce of the pass transistor isn't exceeded.

3) what about that note on the resistance, how should I calculate it? I'm not so good with "sand", you say it should pass Ireq/hFE but what voltage drop I'm looking for?

For Ra, it should be equal to the Vbe of the cutoff transistor. Assume 0.7V if you lack a DMM with diode test.

R = E/I

So let's say 9mA, the middle of your 8-10mA range:

R = 0.7/.009

77.78 ohms.

75 ohms is the closest 5% value.

In practicality, this may be a 68 to 82 ohms, depending on chosen transistors Vbe.


4) what's the math behind the other resistor?

To calculate Rb, you need to know the minimum expected voltage across the device, as well as the hfe of the pass transistor. Say your desired cathode or whatever voltage is 15V, so let's say the minimum is 10V as an example and a transistor hfe of 100.

Ibpass minimum for 9mA is:

Iccs/hfe

9mA/100

Gives us a value of 90uA.

The R is determined by the minimum expected voltage, minus Vbe of the pass transistor, divided by the above result:

(10V - 0.7V)/90uA

9.3/.00009

Gives us 103,333.

100K is the nearest 5% value.

In the interest of stability, you want a wee more juice flowing. Since you know 100K as the maximum resistance, the minimum is limited by the maximum Ib of the pass transistor or the Ic of the cutoff transistor, whichever is lowest.

I's personally drop in a value of 47K in there

 

[SSassen]

Hi Dave, et al,

I've given the suggestions some more thought and I do like the idea of using CCS's in the design. I've drafted up the following schematic in LTspice, would this work? And how do I calculate the proper values for the current through the tube? I plan on using the ECC88.

The mosfet in the cathode follower anode is used as a bootstrap and is AC-coupled with the output. I've used constant current sources in the anode of the voltage follower and in the cathode of the cathode of the cathode follower. Obviously plate/grid-bias voltage is set by two resistors from +VB.

Any suggestions are most welcome.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[sajti]

Hi Dave, et al,

I've given the suggestions some more thought and I do like the idea of using CCS's in the design. I've drafted up the following schematic in LTspice, would this work? And how do I calculate the proper values for the current through the tube? I plan on using the ECC88.

The mosfet in the cathode follower anode is used as a bootstrap and is AC-coupled with the output. I've used constant current sources in the anode of the voltage follower and in the cathode of the cathode of the cathode follower. Obviously plate/grid-bias voltage is set by two resistors from +VB.

Any suggestions are most welcome.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

My experiences, that "simpler is better". Conrad-Johnson made preamplifier with the simple two stage topology. It has very good sound.
I think, that lot of ss devices will results worse sound than the simple valve circuit.
I recommend to remove the gain switch. Simply turn down the volume, if it loud! Set the bias current about 4-5mA, and You will get the best possible sound from ECC82.
But if You can get 5687, or ECC99, You can build single stage preamplifier. This valves easily operate with 10-15mA bias, and You don't need to use follower.

Sajti

 

[SSassen]

Hi Sajti,

The idea is to have excellent linearity and lowest possible distortion and be able to guarantee that without having to rely too much on tube tolerances. If you use anode and cathode resistors you will not be able to achieve that. Total gain is not that important, 4x (12dB) is more than sufficient, but linearity and distortion (as well as noise) are important however.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[sajti]

Hi Sajti,

The idea is to have excellent linearity and lowest possible distortion and be able to guarantee that without having to rely too much on tube tolerances. If you use anode and cathode resistors you will not be able to achieve that. Total gain is not that important, 4x (12dB) is more than sufficient, but linearity and distortion (as well as noise) are important however.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

This topology can gives 0.1% distortion with about 2V output signal. I don't think that You need more linearity...

Sajti

 

[SSassen]

Hi Sajti,

That's too much, with the bootstrapped cathode follower I've shown here you should be able to reach several magnitudes lower than, I do believe that 0.001% is well within reach.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[SSassen]

I guess that instead of a constant current source you can also use current regulating diodes, Vishay/Siliconix CRxxx would be suitable for this purpose as they spec-ed at a maximum of 100V and are available in a broad range of currents? Or would a ring current source with two transistors and a diode yield better results in terms of PSRR, noise or temperature tracking?

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[EC8010]

Either a ring-of-two or a cascode CCS is better and cheaper than a constant current diode (which is only a FET).

 

[sajti]

Hi Sajti,

That's too much, with the bootstrapped cathode follower I've shown here you should be able to reach several magnitudes lower than, I do believe that 0.001% is well within reach.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

OK, I understand. You like numbers, with lot of zeros. I don't really care with the distortion under 1%. Your loudspeaker, and power amplifier create more distortion than the preamp, with the simple topology.
If You check any no feedback, high-end tube preamps, You will find the distortion about 0.1%, and they sound great.
But I wish You to find, what You looking for!

Sajti

 

[SSassen]

With CRD's it is a (much) simpeler circuit, as outlined below. One thing that I need to keep an eye on though is that the voltage across the CRD of the first triode is below 100V, that maximum specified voltage for the CRD.

Comments and suggestions are most welcome!

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[SY]

A few comments:

1. Do you really need a lot of gain in your line stage? For most sources and most power amps, unity is enough, 6dB is more than enough.

2. JFET current sources are bad news. Poor compliance, mediocre source resistance, need for HV JFETs... a cascoded bipolar works MUCH better. I used JFET CRDs years ago, was very unhappy, gave up current sources completely until I was convinced to try a bipolar cascode.

3. Without bootstrapping the ECC88 cathode follower, I had no problem getting the distortion below test residual (0.005% for my tester).

4. The combination of current source plate load plus cathode degeneration works surprisingly poorly. The solution may tie in to comment number 1.

 

[SSassen]

Hi SY,

No, don't need that much gain, 12dB (4x) is plenty, 6dB (2x) more than enough for most of my sources.

The bootstrapping is done for two reasons really, firstly to reduce distortion and secondly to further improve linearity. I have no idea whether this is a necessity, or whether it works just fine without it, I'm just suggesting a few ideas.

How would you go about using a CCS in this particular design? I really would like to continue using it as is, hence with a voltage amplifier and cathode follower, with a ECC88 if needed.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[SY]

Unfortunately, a CCS plate load will force gain to equal mu. So you'll have a line stage with a gain of 30 whether you like it or not.

Still sure you want to use that CCS?

 

[SSassen]

Hi SY,

Alright, so I'm back to my initial schematic I guess? Although I can omit the anode resistor on the 2nd triode if I'm not mistaken, which would result in the design as illustrated below:

Anything I overlooked? Or anything else that can be improved upon? As mentioned I'm using a Velleman K8020 kit as a starting point, hence I have to work within those constraints. For example, a ECC88 can't be plugged in without some modification, as the ECC82 use a 12V6 heater, and the ECC88 a 6V3 one.

Any suggestions are most welcome!

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[SY]

Getting from 12V to 6V may or may not be easy- how is the heater power supply configured?

I'm not sure why that plate resistor appeared in the CF, but I do have a suspicion that it was for stability. With proper layout and bypassing, it should be unneccessary.

Out of curiosity, what's the voltage on the first stage's plate?

 

[SSassen]

Hi SY,

I'm not at home at the moment, so can't measure the voltage, I however do have a schematic of the main power supply if that's of any help? It uses simple Yamicon capacitors and resistors in a number of R/C circuits to do the filtering. By the looks of it this is another candidate for some improvements.

Best regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[kevinkr]

I would strongly recommend getting rid of the gain switching and increasing the plate current considerably from where it currently sits.

Running an ECC82 at increased cathode current will improve its linearity since that seems important to you. I suggest a minimum of 3 - 4mA.. Change that 3.62K cathode resistor to about 1K, leave the 33.2K resistor as is and adjust the series dropping resistor to get about the same plate voltage as you had before. (This determines the headroom and also the operating point of the CF as well.) If you have stability problems due to board layout a small resistor in series between the cathode of the CF to the output coupling cap should help, usually 220 ohms is enough.

I also recommend adding a grid stopper resistor right at the grid of your common cathode amplifier stage again 220 - 1K ought to be fine.

Kevin

edit: fix typo and clarify

 

[SY]

Sander, is this on a PC board or point to point? If the latter, you can put the heaters of two ECC88 in series. In theory, this can be done with a PCB, too, but that's some surgery.

Ditto Kevin's comments about current; 8mA is not too much. Make sure the transformer doesn't get too warm- I don't know what Velleman supplies, but I've seen a lot of smoked power transformers in "classic" preamps where the current was increased. Dynacos were particularly notorious. A quick check of secondary DC resistance would be a good idea.