|
Home
/ FAQ
/ News Classifieds / Events |
Audio Asylum Thread Printer |
Get a view of an entire thread on one page |
173.8.69.177
| '); } else { document.writeln(''); } } else { document.writeln(''); } } else { document.writeln(''); } } // End --> |
In this case, I am trying to determine the optimum Grid Leak Resistor for my 1626 output tube.
Just for clarity, this is the grid-to-common before the grid stopper (if applicable) in a cathode biased RC coupled SET. 1K cathod Resistor.I know that generally for any tube 1 Meg is usually considered max, and I've been reading about grid currents etc, but the spec sheets on the 1626 are vague, and like I said, I'd like to know how to optimize this in conjuction with coupling cap sizing.
I've read the Morgan Jones section on this, but it's not making sense to me, at least not with the specs I can find.
It seems like this is an area that has not been written about much, but is a significant one, and is of value for any amp designer and any tube, especially it's roll in freq and acting as part of a voltage divider?thanks,
robert
BTW, I've been using 270k just because that's what everyone else is using ;)
Edits: 11/02/09 11/02/09
it's really called "grid grounding resistor." a grid leak resistor is the kind that makes use of grid leak to bias the tube so no cathode resistor is needed. those are in the vicinity of 10-50 megohms.
a small leak and a small grid-resistance-to-ground resistor makes the grid set to ground, essentially. hence a grid grounding resistor.
a large leak and a large grid-resistance-to-ground resistor makes the grid more negative than ground, setting up a Vgk bias with respect to grounded cathode. the negative potential that floats there is caused by electrons hitting the grid and not drained away by a little leak, and so voltage starts to settle to a steady DC point until leakage current out grid equals Vg/Rg, and there makes the steady DC bias condition.
to "see" the current flow one must remember that electron flow direction is the opposite of standard positive current flow. So electrons hit grid and leak down grid-to-ground. But the positive direction current flow diagram for a grid leak biased triode is Ip from plate to cathode meeting up with Ig from grid to cathode = total cathode out current to ground.
just had to nit pick.
-kurt
I think it's a misuse of the term unless the tube is being biased by the grid resistor. If it is then it's a 'grid leak resistor'. If it's not then it's a 'grid resistor'.
I have looked at some of the old books and I've been shown to be wrong.
But I still think 'grid resistor' is the proper term. I don't know anyone using "grid leak bias" in modern audio designs.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
We're not using grid leak bias in these designs, so it's not a grid leak resistor. It's just a grid resistor. Asking someone to use the correct terminology when describing a part is not nit-picking.
I want the term "grid grounding resistor" to be used to distinguish itself from both "grid leak resistor" and "grid stopping resistor".
just "grid resistor" is very general. sure it's most commonly used because it is usually not accompanied with a grid stopping resistor.
-kurt
the last tube stage is a grid leak biased one. With Rk=0, minimum Zout is available for drive to the next stage. For it to be good, the CCS is needed for better bias stability.
-kurt
Kurt,
The idea of grid leak biasing the 2nd gain block of a phono preamp is intimately associated with Thorsten Lösch. The rationale for the arrangement is to ease loading of the EQ network and extend bass response.
Search the archives and you'll find a tweaked version of the "classic" RCA circuit I'm associated with. It works fine, without CCS loading the 2nd gain block. MOSFET buffering of the 2nd gain block probably helps in the stability dept., along with improving drive capability.
Eli D.
then i designed it into my phono and Gary copied it, except his CCS is used. That phono stage was designed by many contributors.
-kurt
Yep. I was following Kurt's phono stage development online. Thorsten joined in and suggested using grid leak bias on the output stage.
When I decided to do a phono stage I started with Kurt's design and added my changes which consisted of using the battery bias CCS and using the MU outputs to drive the RIAA network and parafeed output transformers. I wanted to use the drive capabilities of the MU output so the the impedance of the RIAA network could be scaled down ~10x. Lowering the impedance of the RIAA should result in slightly better noise performance and also allowed the use of the Russian Teflon capacitors.
Another change made was change the input tube to the 6FH5. I had more gain than I needed with the Grado cart I'm using. The 6FH5 seemed to clean up the sound overall. I suspect that with the higher mu of the 6ER5 the signal level presented to the 6GK5 output tube was too high for the grid lead bias.
Gary
"I suspect that with the higher mu of the 6ER5 the signal level presented to the 6GK5 output tube was too high for the grid lead bias."
Yes, partially correct there. I measured the overload characteristics of the 6GK5 grid leak stage on the anode. Grid leak bias has two places it operates under: linear, then overload to cause a charge-up on the coupling cap to the grid making it more negative and shifting operating points to a "peak hold" until it discharges back slowly, maintaining still class A linear opertion. This means a very soft overload characteristic and a possible strange shift in sound as it had moved in operating points and continues in another class A area with that operating point drifting back slowly.
If designed correctly, only a short quick overload peak will be nonlinear. But unstable operating points is not the road to hi-fi and it's really important not to overload the 6GK5. One other solution I found was to increase Vgk and bias more deeply by increasing the grid leak resistance, to about 60 Megohms. Then the CCS is the same, but the plate voltage drops into a wider swing area. Vgk is more voltage now that tells you it ought to be in a wider dynamic range area. But you can't measure Vgk in a 60 Megohm circuit. You can only measure the plate voltage here and see that it has dropped, implying Vgk is more negative than before from the plate curves. This was my final solution to that problem.
-kurt
Yes, I read the archives and your and others concerns regarding the term "grid leak resisitor".....
I suggest that in true terms "grid ground reference resistor" (GGRR)is most accurate, since that is what it is doing.....
That is why I clarified in my original post exactly the position I meant:)
Regarding the grid choke, I had previously used it with a hi mu driver, 6n2P-EV (12AX7 equiv) I think, and I really didn't care for it. Too much high end, lost the sweet low bass, no soul.
I don't think that is a result of the choke quality, just my application.
High rp, other issues....
Now I have re-designed several aspects of this SET amp, and the GGRR is one of the last to grasp.....
I find it hard to believe that in the DIY tube group, something as potentially significant as the GGRR hasn't been more fully explored?
I mean this sincerely, with no disrespect intended, as many of you have helped me gain the limited knowedge on this subject that I currently have...... and the ear will always have the final evaluation of "Best"....but I seek the technical background on this also....
peace,
robert
hey-Hey!!!,
So far I've found that lower is better. As low as the preceeding stage is able to drive. If your driver is up to it, I suspect you'll find 200k better than 270, and 100k better than 200.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I'm curious as to why you feel lower is better?
If I'm calculating my freq rolloff right, then 270K and .22uf results in 2.7hz. Lower would require additional uf.
Also, doesn't lower also result in a more attentuated signal to the output tube grid?
robert
Edits: 11/02/09
-----I'm curious as to why you feel lower is better?
Simply because every time I played with it, lower *ALWAYS* sounded better when the preceeding stage drive ability was taken into account.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Ok, thanks.
When I complete this build, I'll play with it lower and see what it sounds like to me;
Actually, when it comes down to the listening evaluations, I also have a couple of MQ grid chokes to give a try...... but that's down the road.....
In the meantime, I'd still like to develop a better understanding of how to optimize this resistor value.
thanks,
robert
Go with the grid choke.
A grid choke gives you the best of both worlds. High impedance so the driver stage is happy and low DCR so the tube's grid leakage won't matter.
You won't be sorry.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Tre',
I previosly did not care for the grid choke, but that was a different driver. My current driver (6GK5) has a much lower rp and may work better.
Since these are monoblocks, and 1 is set up for easier experimentation, I will cobble in the grid choke tonight and give another listen.
But I still want to learn how to calc the optimum grid resistor.....
robert
I don't know if there is a calculation for that.
If you know how much grid leakage current the output tube has you can calculate how much the bias will wander. But tubes will have grid current when over driven as well.
I think you want to use the biggest grid resistor you can (for the sake of the driver) while keeping the bias stable. The problem is, how much of the time does the output tube spend in over drive? And once in over drive, the lowest value grid resistor will be the best (for the output tube, not the driver stage).
BTW Dave has a spice model to show the LF resonance of the tank circuit when cap coupling to a grid choke. The cap value can make or break the overall outcome when using grid chokes.
Let me know and I'll send it to you. Or tell me the inductance and DCR of the choke, cap value and output impedance and I'll post a screen shot of the sim.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Thanks Tre, I think I do understand the potentially very bad runaway issues that can occure with too high a grid resistor and tube is over driven.... grid current flows, plate res goes down, causing more plate current causing more pos grid current etc right?
No bueno....Here is where I am now regarding the grid choke I just installed:
MQ BCP-16 Ni, 2 parallel K40Y9 .22uf coupling caps. I'd like to just use one .22uf......
I think Mike used to say this grid choke is about 4000hy but now just says real high or something like that....and my OPT is also MQ 5k.Right now it all sounds pretty dang fine.
I just listened to Jorma Kaukonen "Blue Country Heart" and Jewel "Pieces of You" followed by Neil Young "Prarie Wind"......
All accoustic, all stuff I listen to and can use to evaluate.It's getting late, but I will just say that it sounds very nice, clear string notes, good bass and vocals are not strident, but carry emotion.
Over all, an initially perceived significant improvement.
More listening will tell.So yes, by all means, I would love to see what you have in a sim.
I have downloaded and played with LT Spice a little, but am nowhere near ready for something like this.
robert
Edits: 11/02/09
I will need to know the dc resistance of the choke and the driver stage output impedance to do the sim.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

Thanks for the help on this.....
Here is a schematic of the driver stage with what I think is the needed info.
Since the 6GK5 driver is listed with 5400 rp and the plate resistor is 15K I assume the output impedance is 3970 ohms.
The MQ inductance spec is kind of vague, Mikey used to call this a 4800hy choke but since that varies with voltage etc I think he said "1000hy" and now just says it's a lot.
I am not including the PS 150hy choke/cap in this calc because I am considering it part of the CLCLCLC driver B+ filter.
We can discuss this more if anyone would like.
These are the current measured specs, the info in brackets is where I plan to go based on more recent fine tuning of the calcs. (no listening tests on those yet)
I look forward to seeing the sim results. I can see that I am seriously going to need to get up to speed with LT Spice soon.....
Robert
![]()
I used the numbers from the website for the choke.
The parallel resistor will dampen this resonance but defeats one of the reasons for using a grid choke, high impedance.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
![]()
Here it is with a .22uf cap.
The peak is sub-sonic either way and not that large.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Your link to the MQ site is different from mine.... good to know that's there.
I assume that your plot represents the resonance with the 20 meg bypass R in place?
What does it look like with just the choke R?
Thanks for the sim, seeing how you represented the circuit helps.
robert
I made the resistor 20meg ohms to take it out of the picture. If I make it 100 megs there is no difference.
If you make it small enough to make a difference, the load presented to the driver will be low. That defeats one of the reasons for using a grid choke.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I have been looking at the periodic A2 recovery from grid chokes lately and have noticed some interesting behaviors.I used basically the same values as you did in your sims but placed the grid choke on the grid of an 801 biased at -30V. I then fed it with a 20V 100hz sine and then stepped the voltage up 6dB for 2 complete cycles and then returned things back to the original voltage.
The sample input signal
![]()
The behavior at the grid of the 801 with a .22 u cap and a 100K grid resistor.
![]()
The 100K resistor replaced with a Grid choke.
![]()
An FFT of the above response to show harmonic content.
![]()
The response in the frequency domain
![]()
Now consider that the 5.5hz resonance is 16dB below the fundamental and is obviously being excited by a pure 100hz tone. Also consider that even though it is 1/7th the magnitude of the fundamental, is is also at 1/18th the frequency which means it produces 2.5X more flux in the core than the fundamental transient that created it. I do want to be clear that this was all the result of temporary A2 operation of an output tube (40V peak hitting a tube biased at -30V)dave
Edits: 11/04/09
Not that I understand what's happening with the grid choke, mine you!
Boats and Fish!
![]()
If it is cap coupled (parafed) then the behavior is the same since the same CL resonance still exists. If it is series fed, then the A2 overload does not cause the same ringing since there is no C for the L to resonate with
dave
Doesn't this series fed power supply have any caps in it? In most power supply designs the last element of the power supply is a relatively large capacitor shunted to ground.Therefore, most of the series type power supplies--- will have all the requisite C's, L's and R's to garner the same potential resonance issues. And may actually be moreso irksome since the whole power supply may have multiple resonances as well as a systemic resonance due to it's relative complexity and the number of reactive elements contained therein.
Paul Joppa in a post several years ago--- pointed this out--- and provided an example to demonstrate that the series fed IT along with the last cap in the power supply--- can also create quite a bit of mischief (i.e., resonance) as I recall from his post.
Ironically-- to my way of thinking--- the parafeed IT will have several less resonant elements in it than the typical series fed power supply/IT combination. In the series fed circuit (depending the on degree of decoupling provided by the last capacitor in the power supply) the signal could "see" several of the reactive elements that make up your series power supply. Gotta remember that (at a minimum) the last cap (which tends towards being large in value) in a series supply is IN the AC signal path.
Builder of MagneQuest™ & Peerless™ transformers since 1989
Edits: 11/06/09 11/06/09 11/07/09 11/07/09 11/07/09 11/07/09
I just didn't know that I did!
Boats and Fish!
![]()
Dave, can we keep our output tube out of A2?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Do you mean for simulations or in reality?
For simulations yes, In reality...I guess. If we consider an amp that hits A2 @ 8W and assume that peaks can be 18dB above the average, we need to be listening to 0.125Wrms to avoid A2 on the peaks.
dave
Hi, Dave!
Boy, do I like your statement! It's just like my measured responses.
I am so happy that you, and not myself, have calculated this RMS figure.
I totally agree with you, and I hope that others will look into Hi-Eff. speaker designs and will work to make them more tonally ideal, since they already are ideal-- mostly, that is, electrically.
Speakers are electric motors. The more power is put into them, the more they "kick back"-- inductively. Our amplifiers view this reverse motion as a chance to slip out-of-time-- musically, and amplifiers also see it as a great opportunity to waste energy.
Kudos to you and your insight.
---Dennis---
Reality.
I clearly need to get some more efficient speakers!
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Or design for proper drive into A2 on peaks. It's worth it IMO.
Naz
Post a Followup: