Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum


Message Sort: Post Order or Asylum Reverse Threaded

LM317 as a CCS , how to?

72.235.7.37

Posted on October 23, 2014 at 19:31:04
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
I was wondering what is the procedure to determine the value of the current setting resistor for the LM317. I have used it on a RH84 ver.2 amp and it seems to improve the Low frequency response. So, I would like to try it on some other tubes, specifically the 7591. TIA, DAK

 

Hide full thread outline!
    ...
RE: LM317 as a CCS , how to?, posted on October 23, 2014 at 19:51:05
Chip647
Audiophile

Posts: 2631
Location: The South
Joined: December 24, 2012

The R is set with the following formula: R = 1.25V / I

So for 25mA 1.25/0.025 = 50 ohms

 

RE: LM317 as a CCS , how to?, posted on October 23, 2014 at 20:31:18
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
Thanx Chip. Is the 1.25 value only true for the LM317? Is there a down side to using the LM317? It seems too good to be true. regards, Dak

 

RE: LM317 as a CCS , how to?, posted on October 23, 2014 at 23:34:07
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
In my case I want to use it set the bias current for a 7591 operating in class A. I am unsure as to what is that value to shoot for. I was checking the Weber bias calculator and it shows the cathode bias at 45ma. Is this the value I should use to calculate the resistor ? regards, Dak

 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 05:00:57
Chip647
Audiophile

Posts: 2631
Location: The South
Joined: December 24, 2012
People generally prefer more elaborate cascode CCS stages then a simple LM317. A simple one using a 10M45 is an improvement over the LM317 is in the hotlink below: implementation of a 10M45 cathode CCS.

This non hotlink link is to a cascode CSS from K and K

http://www.kandkaudio.com/accessories.html

This link is to Tubelab

 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 07:00:36
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
Since you mention the RH84 rev2, I assume you would like to use the 7591 in an SE amplifier. Being SE, the amp can only function in class A.

Dissipation being 19W, and assuming 300V or so across the tube (an application similar to using it in the RH Universal v2 amplifier), I guess 65mA would be a good option (of which probably 58-59mA would be anode current, while the remaining would be g2 current).

From the standard equation posted above in the thread, the current setting resistor for 65mA would be 19.2 ohms. A good practical combination would be 12.4 ohms and 6.8 ohms in series - a DIY friend in the Netherlands uses that combination on his RH Universal with 6P3S (Russian 6L6 equivalent) where maximum dissipation is 18W (similar to 7591).

As for the 1.25V reference voltage, this is a characteristics common to many regulators, ranging from LM317 and LT1085 to TL783 (the latter seems to be referenced with 1.27V but the difference is so subtle that it matters only statistically).
******

http://rh-amps.blogspot.com/

 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 09:41:09
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
Hi Alex, thank you for the info. Do you find any sonic deficiency with using a CCS for setting the bias? It seems to me that compared to the paralleled carbon comp cathode ressistors that it replaced the sound is less smooth. Although the bass is definitely better. cheers, Dak

 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 12:48:48
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
I agree with you regarding the bass improvement, but I have not noticed any midrange degradation.

While the LM317 is most probably not a good option for a current source (above the anode, where it would be directly in the AC path) I consider it an optimal and very simple solution as current sink, i.e. under the cathode.

The fact that it is bypassed leaves it out of the AC path, thus if there is any perceivable change in midrange it is most probably due to the bypass cap. Another issue to consider is the value of this cap, particularly when replacing a previous cathode resistor arrangement. Most often you can find advice to choose very high values for this cap, which is not strictly necessary and depends on the rest of the schematics.

I have started carefully optimizing the cap value from the RH300B onwards. It is thus possible that a value slightly different than given in the schematics of the RH84 rev2 and the RH Universal might be the sonically best option. Anyway, it works fine as per schematics.

Another issue might be whether have you implemented the zener for g2 as well?
******

http://rh-amps.blogspot.com/

 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 13:25:55
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
For my bypass cap I am using a combo of 100uf x 25v Sanyo Oscon paralleled with a 3uf x 50v pio. This was also the cap that was used with the cathode resistor arrangement. So that would seem to rule out the sonic contribution of the caps. I am hoping that it is a break in issue. There is a difference in the current setting resistor for the lm317 which I chose based on the measured current, about 70ma, from the original cathode voltage divided by the cathode resistors previously used which gave me the value of 18 ohms when plugged into the CCS equation. The closest value I could manage was paralleling 3 51 ohm resistor. This arrangement is mostly to see if I was setting up the lm317 correctly.
About the G2 set up; I am using a choke and cap tapped from the B+ to the 7591. I am currently unsure about using a zener diode as many diy'ers feel that there is a noise problem with those. Although I did install them in my rh84 v2 and I find that amp to sound very good. So, I do not know what "zener noise" sounds like. regards, Dak



 

RE: LM317 as a CCS , how to?, posted on October 24, 2014 at 23:20:40
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
Sonic contribution of caps:
I was not referring to the quality of the caps, rather the value. The cathode resistor has a different AC and DC resistance value compared to a LM317, thus bypassing it with the same value cap leads to different frequency response. This should be taken into account from a wider point of view: the schematics of the amp as a whole. Nevertheless, the cap value used with a cathode resistor is most probably adequate for use with the LM317 CCS, although differences in frequency response (and phase) might induce different audible performance in frequency ranges other than bass.

If your caps are new, it might also be a break-in issue. It is not a direct comparison so the results are subjective. You can always try reversing to check whether bass deteriorates and midrange improves - and you might find that the cathode resistor solution is now totally unacceptable overall?

As for paralleling caps, I used to do that a decade ago, because I read how that was a good thing to do - so I just did it. I discovered later that I don't like the sonic results at all, and I would remove the 3uF cap from the eqution. You can try and let us know. It is probably rather large as well. My view is "simple does it".

Zener diode on g2
Zener diodes are labeled as being noisy. This noise can be seen on the scope and probably heard as well in some applications.
Nevertheless, we are using here the zener in series, not in parallel with g2. While the screen grid can be used instead of the control grid to operate a pentode/tetrode tube (I guess it is called inverted triode, with control grid tied to cathode), it is much less sensitive than the control grid, thus eventual noise injected from g2 is highly unlikely to be inaudible. No wonder you did not experience additional noise on your RH84 rev2, I never heard any added noise from any pentode/tetrode when changing from some other g2 arrangement to zener in series.
Thus I can only advice you to do it and check for yourself the quality of the result. Just measure the value of the voltage drop across the choke you are using now (your g2 voltage should be lower than your anode voltage anyway) and apply the appropriate value zener. Remove the cap and the choke from the circuit, leaving just the zener from B+ to g2.
******

http://rh-amps.blogspot.com/

 

RE: LM317 as a CCS , how to?, posted on October 25, 2014 at 13:16:07
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
Hi Chip, you know I have a pair of K&K CCS that I should try out. I bought them for my EICO HF89 project but they were too big to find a place for them under the chassis. I wonder if they would be able to handle the current for the 7591 tube? Also iirc their operating voltage starts at 11vdc. cheers , Dak

 

RE: LM317 as a CCS , how to?, posted on October 25, 2014 at 13:53:10
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
Hi Alex , when I built the amp couple of years ago, I started with the stock circuit and tried different things to improve the sound. Bypassing the electrolytic caps with PIO was a definite improvement in this amp. With the cathode bypass cap the improvement you could say is marginal but I don't feel that it is hurting the sound to warrant removing them which would be a PITA. Right now the amp sounds great with just the LM317 added. It is a definite "thumbs up" change and I will be sure to implement that in my upcoming builds.
With the G2 supply it seems to me that different implementations, only has a subtle effect. When I installed the inductor in the circuit there was a definite improvement from the stock voltage dropping resistor with or without a supply capacitor. But as for the zeners, since they are such as easy way to supply the correct voltage I will be using them in my next build. Thank you very much for the important information and your most helpful suggestions. take care and best regards, Dak

 

Zener Noise, posted on October 25, 2014 at 15:56:24
Triode_Kingdom
Audiophile

Posts: 10011
Location: Central Texas
Joined: September 24, 2006
I've tested a number of Zeners and voltage references in the 2.5V-12V range using an FFT. Some references are actually two-terminal ICs, not a single diode, and those tend to create more noise. The noise varies over frequency, becoming worse close-in. It's not unusual to see 500-1,000 nanovolts RMS below 100 Hz. Low noise Zeners are typically about 100-300 nanovolts. I don't know whether these levels would be audible in vacuum tube audio gear.

 

RE: LM317 as a CCS , how to?, posted on October 25, 2014 at 16:01:40
Triode_Kingdom
Audiophile

Posts: 10011
Location: Central Texas
Joined: September 24, 2006
"Remove the cap and the choke from the circuit, leaving just the zener from B+ to g2."

Is there a reason you wouldn't want to install a cap to ground at the grid, in order to reduce any potential noise and improve short-term regulation? I'm thinking something on the order of 10uF might be beneficial.

 

RE: LM317 as a CCS , how to?, posted on October 26, 2014 at 00:04:20
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
Yes, there are several reasons, that can all be summarized as "superfluous".

The purpose of the cap is to stabilize the voltage and provide an AC path to ground. None of those functions is necessary in this case.

With the CCS under the cathode, current draw is pretty much fixed, and in an SE amplifier this means further guarantee for little or no change in B+.

With a stable B+, and stable/fixed current draw (where the change in current draw for g2 is mere statistics from the zener's point of view), the g2 voltage needs no further stabilization.

The cap is not going to be efficient against potential noise from the zener because the zener might introduce noise that is rather high in frequency for the cap to filter. Caps are usually placed in parallel with zeners in order to try filtering the noise (something you shouldn't do with VR tubes due to the possibility of oscillation), not in series with zeners towards ground.

Last but not least, the cap and the resistor (in a usual arrangement, resistor to g2 and from g2 cap bypass to ground) present an RC filter cell: the higher the resistor value, the lower the cap can be without cutting too much low frequency content. Adding a 10uF cap from g2 to ground will most probably create an inadequate filtering cell.

More often than not, my explanations are misunderstood or not-understood. Most people find it difficult to think outside the envelope, or to think ahead when they want to do something. Zeners are not resistors, and they are solid state technology thus seldom found in old tube technology textbooks. Furthermore, they are usually used as shunt and not series devices. I guess that is the reason why it is difficult to convey those ideas to people with some previous knowledge and experience. But it is easy to check: install the zener, as advised - and check it out. OK, now you can experiment by adding the cap. Are you going to like the result? That's up to you to decide: I guess you are not going to like the result, but it seems that unless you do it, you are going to be left with doubts on your mind.
******

http://rh-amps.blogspot.com/

 

RE: LM317 as a CCS , how to?, posted on October 26, 2014 at 01:28:54
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
If one wants to set a constant current for a tube, and the CCS is going to be under the cathode, in my view there is going to be little difference between a "simple" (the circuitry inside is actually not so simple at all) IC like LM317, and some "elaborate" (actually simpler than the IC in terms of number of elements, most probably) discrete components solution - is going to be marginal, if any. Likewise, some claim being able to hear the difference between LM317 and TL783 in that position, but to me it is marginal if not placebo.

Besides the simplicity of implementation, parts count, and reliability, what matters is the voltage drop. The LM317 requires 1.25V for current setting and probably at least 2.5V across the regulator, thus a minimum cathode voltage is going to be 3.75V for correct operation.

While some "fancy" CCS made with discrete components might be applicable as current source above the anode, where the voltage drop is not an issue, as current sink with tubes that have rather low cathode bias voltages, in the range 5V to 10V, most of those solutions are not applicable.

Simplicity is also an important part of the equation: 1 IC with a series of protective features plus 1 resistor makes for easy installation and probably higher reliability, not to mention the cost...
******

http://rh-amps.blogspot.com/

 

Using a 6SL7 instead of 12AT7 in a RH amp, posted on October 26, 2014 at 11:10:31
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
This is way off topic but since you are actively following this topic I hope you don't mind if I tap into your experience for some advice.
My next build will be a RHKT88 amp with a 6SL7 driver and I just wanted to know what would be changes needed to use that tube instead of the 12at7. My output trans is about 20 watts capable of 2.4K primary impedance into 16ohms. Regards, Dak

 

RE: Using a 6SL7 instead of 12AT7 in a RH amp, posted on October 26, 2014 at 13:34:36
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
If you build an RH Universal v2, you can choose to install an octal socket instead of a noval socket for the driver. This will obviously limit the choice of driver tube to 6SL7 and 6SN7, plus the various ECC3x versions.

While the v2 circuitry is optimized for ECC81 (100k Rfb, 2nd 100k in series shorted by switch), and ECC83 (switch open, both 100k resistors in series for Rfb), you can use other tubes that fit the socket pin layout.

For 6SL7 and similar, the 200k position is fine. For 6SN7 you can try both switch positions and see which one you prefer. You can see on my blog the RH Universal v2 operated with a 6SL7 near equivalent (ECC35 if I recall correctly) using an adapter.

The OPT sre going to be fine if your speakers are 16 ohm nominal. For the rest, stick to the schematics including the power supply if you want to have all the output tube universality.

And, once you do, let us know the result!
******

http://rh-amps.blogspot.com/

 

RE: Using a 6SL7 instead of 12AT7 in a RH amp, posted on October 26, 2014 at 14:13:37
DAK
Audiophile

Posts: 2712
Location: PACIFIC
Joined: August 8, 2010
Hi Alex and thanx for the help. I found that trying different values of the fb resistor in my other SE pentode amps can help with fine tuning the sound. For example to little fb makes the sound softer and the bass bloated. Does changing the fb resistor values in the RH amps have a similar effect? I am thinking that since the 6sl7 mu lies midway between the high mu of a 12ax7 and medium of the 12at7 It may be beneficial to use a fb resistor somewhere between 100 and 200k ? regards, Dak

 

RE: Using a 6SL7 instead of 12AT7 in a RH amp, posted on October 27, 2014 at 04:26:40
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
DAK,

I guess you are referring to "overall feedback" across the amp (from transformer output back to input).

Fine-tuning the Rfb in RH amps will not give the same result. While in simplified theory it may come to "softer sound, bloated bass", it is actually much more complicated than that.

With direct feedback from the anode of the output tube to the anode of the driver tube, there are several issues to take into account, like:
- internal resistance of driver tube,
- mu and gm of driver tube,
- anode resistor as a means of determining the gain of the driver tube,
- feedback resistor as a means of determining feedback, but also as in DC terms as current flow path,
- cathode resistor of the driver tube, as a means of defining gain and as well feedback path, etc.

Thus the 100k option is optimized for ECC81, while the 200k option is optimized for ECC83. When I state "optimized" I refer to "maximum power output with lowest (or a pre-defined target) distortion".

Other tubes, like ECC82, 5751, E180CC, 5965... can be used as well with one of the settings (we cannot have miriads of settings, you know) with rather good results, although the settings are not what would be optimal for a given tube. On the other hand, it is not all that important whether you otuput 10W or 12W, the difference is rather small in terms of loudness - and if you like the sound, there is not reason why not.

Now, using the ECC81 with 200k will not lead to "bloated bass" or something similar - but the sound is going to be different and the output power not so optimal. Using the ECC83 with 100k, on the other hand, is going to give probably no more than 5W of output power... because at that setting the ECC83 just cannot drive the "output circuit" (the output tube and the feedback loop) any better than that: it needs "help" in the form of adjusted feedback.

That said, the 200k setting is not optimal for 6SL7, but it will do very nicely! I was very pleased when testing it with both 6SL7 and ECC35, although I did prefer the ECC35 - but because my 6SL7s are "garden variety" while the ECC35 has a much higher "intrinsic sonic value".
******

http://rh-amps.blogspot.com/

 

The Cap is not Superflous, posted on October 27, 2014 at 11:48:18
Triode_Kingdom
Audiophile

Posts: 10011
Location: Central Texas
Joined: September 24, 2006
"The cap is not going to be efficient against potential noise from the zener because the zener might introduce noise that is rather high in frequency for the cap to filter."

I don't agree. Zener noise decreases with frequency, whereas the XC of the cap goes down. A single cap can easily reduce Zener noise by a factor of ten, and it's performance does not suffer at greater offsets. If anything, the difficulty lies in attenuating noise close-in (low frequency). I'm very familiar with this as part of my day job.

 

RE: The Cap is not Superflous, posted on October 27, 2014 at 12:11:46
Alex Kitic
Audiophile

Posts: 182
Location: Serbia
Joined: June 25, 2014
While I remain with what I already stated - just like I stated, I have nothing against experiments and real-life trials of hypothesized solutions.

In this case, I guess it is rather easy to connect g2 to B+ via the zener diode, and try it as it is. After that, a cap can be added from g2 to ground, and the result auditioned: if one finds that is sounds better with the added cap, or that there is less noise -- I really cannot object to the solution being implemented.

From my experience, there is no relevant noise to start with - which might have been contributed by the zener diode. Once there is no noise to cancel or filter - the cap becomes redundant, or superfluous, you are free to choose the term...

Whoever wants to try it for himself, should do it, and let us know :)
******

http://rh-amps.blogspot.com/

 

RE: The Cap is not Superflous, posted on October 27, 2014 at 13:25:22
Triode_Kingdom
Audiophile

Posts: 10011
Location: Central Texas
Joined: September 24, 2006
"From my experience, there is no relevant noise to start with - which might have been contributed by the zener diode."

Yes, it may very well be that noise contributed by the Zener is insignificant in the larger picture.

"Whoever wants to try it for himself, should do it, and let us know :)"

This is a place for measurement first, listening later. The question is whether Zener noise contributes to system idle noise at the output of the amplifier. Whether a particular owner can hear it with a given system is irrelevant to optimizing the design of the amplifier. A FFT (hardware or software) is required to analyze this issue accurately.

Also, just to address one other point you raised to support the claim that the cap is superflous...

"Adding a 10uF cap from g2 to ground will most probably create an inadequate filtering cell."

The 6BQ5 G2 draws approximately 6mA. That current through the Zener, along with a 21V drop, equates to an effective impedance of 3,500 ohms. Under these conditions, a 10uF cap at the screen will decouple the grid sufficiently to create more than 45 dB attenuation at 1K. Whether this is effective at the output depends on many factors, including the AC impedance of the primary B+ supply. At the very least, I would consider this a good safeguard against degradation (likely in the form of distortion) by a screen that that might not otherwise be sufficiently regulated.

 

Page processed in 0.024 seconds.