How to make an 600 ohm LCR have Zin=1200ohm

the road not taken.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Thanks Dave,

Think you got it all covered. Nickel seems to be the way to go.

I have my TVC and feel forced to make use of that investment.

As I am no tranny-man, please explain to me what happens with Zo, with respect to the involved DCRs, in the attached file. Not sure if it will even work. We assume the autoformer is a 1:1 100H with 500ohm DCR/winding. Maybe a DC-free autoformer could be made with lower DCR?

My guess: With the assumtion of DCR of 2*500 in the autoformer, Ri+536+666=3k, 3k/4=750, 750+500/4+500=1,33k, 30k//1,33k=1,3k

This leads me to the fact that I have to find a way to make the next stage accept the 3k directly from the bifilar 8) .

But I am not sure this is the way to go as the 1000u cathode resistor might degrade the whole setup :cry: .
Attachments
reVRIAAwhitebifi3.asc
(7.42 KiB) Downloaded 796 times
Brgds
Lars
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

dave slagle wrote:hey guys,

thanks for the Z plots, i now know how to do it in spice (forgot you could enter an expression)

i never realized that the unterminated LCR still provided roughly the same load to the driver.

thanks for the education.

dave
Okay, so let me get this straight, if you drive the normal 600 ohm LCR with a source impedance of 600 ohms, it presents a load of 1.1K to 1.9K to your source?

Also, if you choose to terminate the LCR with a 600 ohm resistor on its output side, it presents your source with a 600 ohm load?

I wanted to make sure I fully understand the above. You really need to know the load that the LCR network will present to your source. If you want a White cathode follower to drive it, you have to optimize the top tube's anode resistor for balanced current for your load.

Rgs
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi,

Your RIAA must be terminated with 600R and loaded with high Z ..OR.. terminated with high Z and loaded with 600R.

In the later case, the White Cathode Follower's total Zout must be 600R. You could achieve that loading several ways, either set the various load resistors in your designto give exactly 600R, or close as feasible, or make your design for the lowest Z out you can obtain and add series resistance to make it up to 600R, which is what Lars does.

Hope this helps.

-- Andrew
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

Andrew wrote:Hi,

Your RIAA must be terminated with 600R and loaded with high Z ..OR.. terminated with high Z and loaded with 600R.

In the later case, the White Cathode Follower's total Zout must be 600R. You could achieve that loading several ways, either set the various load resistors in your designto give exactly 600R, or close as feasible, or make your design for the lowest Z out you can obtain and add series resistance to make it up to 600R, which is what Lars does.

Hope this helps.

-- Andrew
Hi Andrew,

I understand the loading and/or drive requirements of the 600 ohm LCR network. I want to drive the LCR network with a 600 ohm source so I can terminate its output anyway I want. I just didn't completely understand what the impedance is looking into the network. I was under the impression that the LCR network looked like a constant 600 ohm load. But, on the second page of this thread, some impedance graphs show maybe otherwise. In order to optimize the current balance of the White CF, you need to pick the right resistance for the top tube's anode. The required resistance is dependant on the load you what to drive with the White CF. This load would be the impedance of the LCR network, plus the series resistor used to create your 600 ohm source impedance.

Rgs
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

I'm not completely familiar with the White CF, can you provide more text or a link describing how to optimize the anode R?

As far as I can tell, if your loading with a matched value, and terminating with high Z, and looking at the graphs I have, then the load looking into the LCR does indeed vary, but it is pretty much a constant once you hit roughly 200Hz.

At the all important 20Hz with the Tango I get 380R, at 50Hz its 420R and by 100Hz its 500R, at 200Hz its up to 570R, rapidly hitting 604R thereafter.

I'd be very keen to hear more if its possible to design a driver stage that had the same characteristic Z as the load looking into the LCR?

-- Andrew
Attachments
EQ-600test.txt
Stick this lot in a spreadsheet and ask for a plot to see the how the load varies when presented with a constant 600R source.
(7.26 KiB) Downloaded 666 times
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

Andrew wrote:I'm not completely familiar with the White CF, can you provide more text or a link describing how to optimize the anode R?

As far as I can tell, if your loading with a matched value, and terminating with high Z, and looking at the graphs I have, then the load looking into the LCR does indeed vary, but it is pretty much a constant once you hit roughly 200Hz.

At the all important 20Hz with the Tango I get 380R, at 50Hz its 420R and by 100Hz its 500R, at 200Hz its up to 570R, rapidly hitting 604R thereafter.

I'd be very keen to hear more if its possible to design a driver stage that had the same characteristic Z as the load looking into the LCR?

-- Andrew
Andrew,

Thanks for the text file. This will help me. One of the best papers on the White CF can be found here--> www.cavalliaudio.com/cj/docs/WCFOptimization.pdf

I had to read it several times before I fully understood the concepts in it. The thing I found very interesting is you have to consider both the small signal and large signal conditions for the White CF. There is no perfect anode resistor, but there is a best compromise based on the load you want to drive. The thing that makes the White CF so great is it can have symmetrical current drive if setup correctly. This is something a normal CF can't do. In addition, the White CF has lower output impedance and can deliver twice the current to the load as a normal CF.

I made up an Excel worksheet to calculate all the parameters of the White CF. The formulas are from Alex Cavalli's paper and the Tube Cad Journal.

A few notes explaining my worksheet. The Blue colored cells are for inputting values. The red cells are the calculated results and should not be changed. You input Rp, Mu, Bias voltage, and the bottom tubes Rk value at your desired operating point. It is important not to just pluck the normal operating values from the tube data sheet. You must enter the values at your operating point to get accurate calculations.

I also have an input for the desired load to drive, and a minimum and maximum load resistance. This allows calculation for how much current imbalance the varying load will cause in the White CF. A current balance of -1.0000 is perfect symmetry, and a current balance of -1.0400 or -0.9600 is a 4% imbalance of current. Of course this is at maximum output voltage, which we should never approach with a good design.

At Line A20 is the ideal small signal upper tube anode resistor value. Just below it is a manual input value. I did this because I wanted to be able to see what other values of Ra would do to the other calculations. Down at line A41 is the calculated ideal large signal Ra value. If you input this value at A21, it should provide the best large signal symmetry for difficult loads. However, I have found that a value somewhere between the calculation at A20 & A41 to be more practical.

At Line A24 is the calculated gain of the White CF. Of course it will always be less than 1. In addition, I calculate the maximum input signal before you drive the grid of the bottom tube positive. This is found at line A32. I also calculated what the maximum output voltage should be. All the calculations don’t always line up. It is obvious if your gain is less than unity it is impossible to get more a voltage swing than what the maximum input is. So, line A30 adjusts the maximum output to correlate with the small signal gain and maximum input allowable.

Part of calculating the current imbalance has to do with the expected linear operating range of your tube. For this, I have lines A35 & A38. You input the percentage of range you expect the tube to operate in a fairly linear fashion. I’m a bit conservative, so I plugged in 70% and 75% for my example. You can change this to what ever fits your design best. I just wanted to be honest with myself and not over predict what the results would be. Well, I guess that is it for now.
Attachments
White cathode follower.xls
(22 KiB) Downloaded 635 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hi Andrew and Jeepman,
Have not been at the forum for a while but read the last of your posts this morning.

The Excel spreadsheet seems to to be exactly like it should, great work. I used two of Broskies articles to predict the top resistor. Will read the interesting Cavalli-paper after work today.

I used Ltspice to get in the ballpark of the "series" resistor and the top resistor.

When the RIAA is readybuilt my intention is to use the series resistor for finetuning of the bassresponse and adjust the topresistor to get maximum swing with least distorsion.
Brgds
Lars
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Yes, I need to read the article too, I'm afraid I'm still following the high gm pentode route into the 1k5 RIAA, but the White CF seems an interesting approach, and I too have some reading to do.

I'd very very interested in hearing the results, especially with respect to self noise, having 3 valves on the input, assuming noise is not a problem this could be an interesting approach.

Jeepman, what valve will you use?


-- Andrew
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

I'm looking at using a 6M11 tube for the front end. This is a high Gm pentode with two identical high Gm triodes.

Data sheet here---> http://www.mif.pg.gda.pl/homepages/fran ... 6/6M11.pdf


Image

The pentode will be ran at zero bias with a screen voltage of 50V. The most positive the grid will be driven at phono levels is around 50mV. Even at this, it doesn't draw enough grid current to be a problem and I get to rid myself from a cathode resistor and cap. The gain of the pentode stage becaomes Gm times the anode resistor. This will direct couple to the White CF made of the two triodes. These will be ran at 125V, 11mA, -0.68V. This will net a Mu of 60.8, Rp of 5800 ohms, and Gm of 10.5mA/V. The White CF will be able to swing over 11 volts, so no problems driving the 600 ohm LCR network.
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

Andrew wrote:Yes, I need to read the article too, I'm afraid I'm still following the high gm pentode route into the 1k5 RIAA, but the White CF seems an interesting approach, and I too have some reading to do.

I'd very very interested in hearing the results, especially with respect to self noise, having 3 valves on the input, assuming noise is not a problem this could be an interesting approach.

Jeepman, what valve will you use?


-- Andrew
Andrew,

I plan to build and listen to both a White CF and the high Gm pentode route. The one thing I really don't like about the White CF it is very sensitive to changes in load impedance. Once you optimize it for one load impedance it doesn't take too much variance to imbalance the two triodes. However, I really like the very low output impedance and close to unity gain. I also like the power supply rejection ratio (PSRR) is very high. I don't think the White CF would add any significant noise. Of course some of the advantages of the high Gm pentode route is good PSRR and easy gain calculations. Gain is just Gm times the anode resistor. Also, since the anode resistor will be around 600 ohms (for the 600 ohm LCR module) tube ageing becomes a non-issue and can’t affect the LCR network. It is not an easy choice. There are pluses and minuses for both circuits. Like I said, I’m just going to build both and listen and see which one sounds the best.
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Interesting valve choice, not met that one before, seems perfect for the job in hand.

Just for interest, a friend of mine and I both tried zero bias, he found it caused problems on some step-ups the grid current and DC where is wasn't welcome etc.

I found clicks and pops were more obvious so I assumed I was getting overloads, that was on a D3a and E810F. Other valves, of course, will be different.

In the end I found a red LED was a better bet and spent some time scouring the data sheets for one with a nice linear impedance, it turns out the same LED is used by a kit manufacturer for the same reason.

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »


Gain is just Gm times the anode resistor.
Just another thought, bit of a brain dump, sorry......

Anyway, I was worried about the 600R load on the pentode, since gain is driven load parallel with the load resistor so for 600R LCR overall gain becomes gm * 300R (thanks to Lars for setting my theory straight on this) Anyway, this is why I got Dave to wind me chokes for 1500R. Getting it all spot-on took a good deal of effort and should thank Dave again for helping to get this right, not simple designing an LCR even when starting from Tango's legacy.

-- Andrew
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Jeepman:
Have seen the name on the spreadsheet somewhere else. Was it you who had a Tractrix discussion with my swedish forum-friend Ronnie Eriksson at the Romy-site?

Anyway, tried the spreadsheet and got a somewhat smaller topresistor-value than I had calculated with my 6C45 WF.

Will post my new schematic later as I saw the one in my last post was not up to date. The new one is with two bifilarcoupled gain-stages before the LCR and direct out from the LCR with about 300mV output level.

EDIT: New schematic

Image

To clarify things, the White-follower will see a load from 980 to 1204 ohms in this circuit. A quick-test in the spreadsheet indicates a Ra just below 100ohms.
Brgds
Lars
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

reVintage wrote:Jeepman:
Have seen the name on the spreadsheet somewhere else. Was it you who had a Tractrix discussion with my swedish forum-friend Ronnie Eriksson at the Romy-site?
That would be me.

Rgs
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

Image

Well, as promised (a very long time ago) I muttled my way through several circuits to arrive at the one above. As best as I tried, I could not get the White Follower to sound as good as this one. To my ears, even when balanced to the extreme, the White follower did not sound as focused. My guess it is the whole push-pull thing going on again. Damn these single ended circuits, they just wont quit! :)

Due to more grid current issues than expected, I had to place some small value resistors in the cathodes. As far as I'm concerned, I'd still call them zero bias because the values are so small.

I'll be in and out, so any follow up on this post may take a while. However, I'll try to check back in to answer any questions.
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hi Jeepman,
Have to check with you. With an Iq of 35mA the available output swing will be 3-4V rms and with an assumed Gm of just over 30ma/V(from Millets pentode survey) the gain will be ca 10x.

The overload margin will be OK but do you really consider the gain before eq to be high enough?
Brgds
Lars
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

reVintage wrote:Hi Jeepman,
Have to check with you. With an Iq of 35mA the available output swing will be 3-4V rms and with an assumed Gm of just over 30ma/V(from Millets pentode survey) the gain will be ca 10x.

The overload margin will be OK but do you really consider the gain before eq to be high enough?
That's what a input transformer is for. :D

I measured a gain of 13.8 for the 12GN7 stage going into the LCR module. Gain for the 6HB6 is about 172. Then there is the 20dB drop across the LCR module. with a 5mV 1KHz signal, I get 1.2V out. So overall gain is about 239X or 47.5dB. This is enough gain for my system. However, it sounds better with a 1:5 input transformer.

Rgs
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Jeepman,

Great stuff, I have some 12GN7 looking for a home, perhaps I should try them.

What does the Tango circuit look like internally, does it offer a DC path to ground?

I had a DC current when the cap was after the RIAA module and the module was at the anode DC of the input pentode. My modules were passing a couple of milliamps, perhaps more, across the 'internal' chokes and down through their ground connection. Dave wasn't certain if this would cause a problem but I felt things were better with no DC current through the chokes.

Just a suggestion, but the input pentode will offer more gain if you bypass the cathode with a cap; a BG could be good here. I would have said the extra gain before the RIAA is worth the cap, especially in terms of dynamics, but its your personal preference, of course. I ended up using diodes, but I looked for very linear ones. I have some BGs on loan to try out.

Mills seems the preferred choice for anode loads.

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

the EQ-600P will indeed pass some DC. With 150V across it and no terminating resistor I get 6.5ma through the first choke and 1ma through the second.

dave
Attachments
eqp600p.gif
eqp600p.gif (86.27 KiB) Viewed 33506 times
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hey Dave,

Thanks for the circuit.

Yes, that would tally with what I saw, I got a bit less as the chokes you wound would have been higher DCR, I guess?

Do you think that DC current would make a difference in the Tango's case?

Jeepman, are the Tango's potted? If not, you can stop the DC path by putting some caps on the 14k and 68k.

-- Andrew
Jeepman
Posts: 16
Joined: Wed Mar 05, 2008 2:17 am

Post by Jeepman »

Hey Dave,

Thanks for the circuit.

Yes, that would tally with what I saw, I got a bit less as the chokes you wound would have been higher DCR, I guess?

Do you think that DC current would make a difference in the Tango's case?

Jeepman, are the Tango's potted? If not, you can stop the DC path by putting some caps on the 14k and 68k.

-- Andrew
Andrew,

I'm not 100% certain, but I believe the Silk brand LCR modules use a modified layout that only needs DC blocked on the input or output. I never tried to measure DC current on it. However, when I tried to measure the DCR on the input, my meter behaved like I was measuring a capacitance. See below for what I believe is inside my modules. And yes they are potted so I can look inside.

Image
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Jeepman, as far as I know that's the same circuit as used by S&B in their modules, I measured a pair of those and they looked like there was no 50Khz point which is one place where they would pass DC.

So, it seems you're clear and there's no DC path to ground.

If you want to be doubly certain, and can't see inside, I would put a small resistor in series in the ground connection and see if there's any DC voltage across the resistor.

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

somewhere I saw a reference to the various LCR topologies and the "Silk" version was the original and tango added the resistors to increase the accuracy. I wanted to add the third (allen wright) time constant to eliminate the droop at 20K and thomas Mayer helped me out with that.

While the sims are great, the two things that i found to be very important when attempting to get the inductors to behave as the sims are the linearity wrt frequency and the capacitance of the coils. Since my version was about 12K the value of the 600R versions this became quite a task but my end results measured good out to 40K before the capacitances showed up. I did note that in the versions without the 3rd time constant added the capacitances could be used to "fake" it but that resulted in the resonant peak showing up much lower in frequency.

dave
Attachments
the measured results of the two channels against the normal (blue dashed) and 3rd time constant (orange dashed) ideal response
the measured results of the two channels against the normal (blue dashed) and 3rd time constant (orange dashed) ideal response
Picture 7.png (34.77 KiB) Viewed 33527 times
simmed response of the three networks.  the tango and jeepman are both 600R and mine is 7K30 (I could easily scale mine back to 600R)  Also note the phase response :-)
simmed response of the three networks. the tango and jeepman are both 600R and mine is 7K30 (I could easily scale mine back to 600R) Also note the phase response :-)
Picture 6.png (18.95 KiB) Viewed 33527 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

It is easy to get rid of the DC path through the inductors when you use the EQ-600 schematic. . Just connect the 14k and 68k with a 4,7UF to ground like I did.

By adding a resistor between the connection-point of the two 600ohms and the ,126u cap you will be able to get the third time constant. I simmed it to 51ohm.

You can see the modifications in my schematic in a post above.

I just sold one of my 4 Tango EQ-2L pairs via E-bay and have now ordered a pair of Lundahls to try out. There is one inductor in every container. That should make it easy to make a better pcb-layout.
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Did a sim with a 0ohm Zout inverse RIAA with the 3rd timeconstant directly feeding via 600 ohm serial resistors, output load 30k . My modified was with a 56ohm instead of 51 ohm earlier mentioned. Would have liked to incorporate Daves 7k but can not find it. Maybe Dave could help?
Attachments
seri60030kout.JPG
seri60030kout.JPG (91.52 KiB) Viewed 33388 times
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Did some poking around with the Tango values. Zin=600ohm, Zload=30k. Will work a little more on it to only use standard component values.
Attachments
LCR.PNG
LCR.PNG (35.16 KiB) Viewed 33373 times
LCRjfr.asc
(8.55 KiB) Downloaded 454 times
Brgds
Lars
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

+/- 0,017dB using standardvalues.
Attachments
UltimateLCR.asc
(6.14 KiB) Downloaded 525 times
Tangorev600in10kout.PNG
Tangorev600in10kout.PNG (36.6 KiB) Viewed 33350 times
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey lars,

My resistor for the 3rd TC was in the 500-600 range (close enough to 10X that of the 600R unit) and needed to be adjusted in circuit to take into consideration the capacitances of the inductors in question. Too much capacitance in the first choke and the response rises and too much capacitance in the second choke and it falls. The two can actually be set to offset each other and flat response seen without the additional resistor.

The biggest problem i encountered was the linearity of the inductors WRT frequency which is why I feel they all should be tapped to allow for in circuit adjustment to easily adjust for their inherent flaws to select the best compromise for the given situation. If you think about it the only real variable when it comes to getting this circuit to behave as simmed in this situation is the behavior of the inductors. (and the ability to measure with anywhere near the accuracy spice magically comes up with)

I'll attach the .asc from the plots above. My choice of impedance was based solely on easily obtainable caps and not some moronic number from the past. The ratio of the two caps is fixed and at the ratio of 40:1 so anyone that tries to replicate these circuits would do themselves a favor and take heed of the easily available high quality caps. For me .4 and .01 were simple which suggests that all the panic of making a 5.02 and a .126 could be alleviated by saying 730 ohms is OK and a 4u and a .1u may actually exist. Of course then given the wonders of spice, when the two caps in question are obtained and their value verified to the needed precision, the rest of the easily verifiable values (resistors) can be identified and the only variable becomes the inductors. They can be tapped and the capacitances can be reasonably accounted for by the use of the ~51R resistor (measured in circuit to the needed precision) :-)

dave
Attachments
tango jeepman slagle LCR.asc
(10.91 KiB) Downloaded 573 times
reVintage
Posts: 365
Joined: Fri Sep 07, 2007 9:54 am
Location: Sweden

Post by reVintage »

Hey Dave,
Interesting about the 7k, will try the sim. But you should use my inverse RIAA instead as I think it is better. It is also based on the Hagtech paper, see my attachment. Is it possible to make an CF inductor as shown with a DCR target of 320-340ohm?

I also used available component values as this must be the way to go, even if I already have the inductors(and I have only measured DCR). My values as well as yours are based on theoretical spice sims.

Most of my final adjustments can be done by changing the 600ohm input(the Zout of the presceding stage must also be substracted from it) and the 3rd time constant 56 ohm. The Miller capacitance of the tubes used will also make the 56ohm value to change.

It is important always start with the best theoretical model and then fine adjust from that.

About the 7k, it should be more sensitive to capacitances in the inductors etc?
Attachments
TangoLCR6H30CF.asc
(7.38 KiB) Downloaded 520 times
Brgds
Lars
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey lars,

A 25hy 40ma 330 ohm inductor is pretty simple see the plots below for L vs. I.

Indeed the capacitances are more of an issue with the 7K mainly because the chokes are bigger. It would be interesting to compare the CF approach of the parallel 6H30 to a version where the 6H30's are used as plate followers into a series resistor at say 4X Rp then into an unterminated LCR at 5X Rp.

dave
Attachments
plots from a 49% nickel choke.
plots from a 49% nickel choke.
Picture 1.png (29.11 KiB) Viewed 33290 times
Post Reply