• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Unity coupled circuit destroying tubes/valves?

Besides the incredible engineering skill Norman had, he also had one other thing that I think was truly unmatched in the his field, and it put him head and shoulders above all others: His communicative skills. He had an ability to write in such a way as to make the complex understandable like no other. I too have many of his works, and consider them some of my most prized pieces. Any library is expanded multifold when his work is added to it.

Dave
 
The first Mac related article is the one I mentioned earlier, while the second one is the one they reference in it. Thanks for posting it!

Dave
 
A great overall resource,already mentioned many times I'm sure, especially if you're looking for articles in old electronics magazines,can be found here:

americanradiohistory.com

Be warned though,I have found that using this site has considerably reduced the memory space on my computer,considerably increased my budget for printer paper,
and sleep has become a thing of the past...........
 
Indeed a very interesting solution to the notch distortion problem -- although overall harmonic distortion would still be on the order of that which a conventional pentode, triode, or UL stage produces as the case may be. The Unity Couple output stage does in fact have severe drive requirements, but returns very low overall distortion in the process, and even lower output impedance (since the FB employed is effective in both directions of signal swing), including the elimination of the notch. While interesting no doubt, advancements in winding conventional output transformers largely eliminated the need for such a measure. Thanks for posting that information!

Dave
 
Hi everybody.
I just built a 60 W amplifier derived from MC60 schematic by using a Vanderveen toroïdal transformer. I have used a VDV-1070-UC. The coupling is very good between the taps (But take care to the fact that the tap 1-2 is coupled with the tap 4-5 and tap 2-3 with 6-7). The measure of DC resistance allows to check (same resistance : same winding location)). If you cross the connection by using for example tap 1-2 with 2-3, the amplifier works but the answer on square signal is not perfectly balanced)

I have made some changes in the design of the circuit :
- Power supply uses diodes instead valves.
- The bias circuit used two diodes instead one selenium rectifier.

I got some oscillations at the beginning. After investigation, I found two ways for solving that :

1- adding a capacitor of 47pF between the pin 1 and 2 of the 12AU7, and adding series resistor of 2.7 K on the grid of 6550. The bandwith in this configuration was limited to 70 kHz at -3dB. No overshoot on square signal.(resistor set à 2.15 kohm in the feeback instead 1.3 k)
2-no additionnal capacitor nor resistor. I have just decreased the capacitor in the feedback from 470 pF to 69pf (= 47 +22) . The bandwidth is increased to 410 kHz ...at -3 dB. I have two areas in the bandwith with a gain increase (ringing at 153 kHz : +0.4 dB and 290 kHz : + 1.6 dB). The amplifier remains stable but, I have overshoot on square signal. The increase of the capacitor in the feedback (I.e : 94 pF) leads to unstability.

I have kept the amplifier in the configuration 2. In the audio band, the overshoot is not a constraint. It sounds pretty well . The rise time is impressive and much more better than in the conf 1.

At the time being, I try to reproduce exactly the behaviout of the amplifier in LTspice for improving the answer on square signal.
 
Last edited:
May we assume that all of these results are based on using a non-inductive resistive load for testing? I would think that configuration 2 would be much less stable into either no load, a real world speaker load, or worse yet, a capacitive only load -- and leave the amplifier susceptible to every little transient that comes along as well with such great bandwidth. The answer would appear to be found in the approach used in configuration 1 -- reducing amplifier response locally, as trying to do so through the global loop brings on the instability.

Dave
 
The test has been made with a resistive load having a small inductive load. No issues on two kind of speakers, but not very capacitive. But, yes, i agree, the conf 1 is certainly more safe.
 
Interesting...if I understand correctly, albalkar, you cloned the MC circuit identically and connected the Vanderveen VDV-1070-UC as though it were the McIntosh output transformer?

Are you using 6550 output tubes and what power output are you getting?
 
Roughly. I have made some modifications in the original circuit (diodes instead valves in the power supply. The bias circuit has been slightly modified due to the change of the power supply transformer, feeback resistance adjusted also).I have also put the power supply in a separate box. It is very efficient for the noise, I have a level of noise below my MC30 (~ 1 mv peak on the output). I got a little bit more than 60 Weff into 5 ohms.

I did that because, for me, it is almost impossible to find a genuine MC60. And I have speakers with high efficiency. My purpose is to obtain a higher signal/noise ratio. It would be a pity to customize a genuine MC60, It is not an issue with a DIY amplifier...;) Today it works properly. (I had a comment on my previous post : the choice 1 works if the tap 4+5 and 6+7 are crossed.). .
Initialy i tried to build an output transformer and not using the 1070-UC. I have met some issue. The insultation was not great and the transformer has blown up...So I have used the VDV. There is three drawbacks :
- This OTP is sensitive to the bias balance between the 6550. the valves I use have been matched.
- The power in 8 ohms is below 60 W (a bit above 40 Weff)
- The higher bandwidth of the OTP request to performe some adjustements and it seems difficult to have a perfect behaviour on square signal.

But it works and sounds great. I have tried the amplifier on two kinds of speakers w/o issue of stability.

The next steps are :
-to continue to work for winding an output transformer with the right characteristics. (My final purpose is to build a version with a double push-pull of 6550 of 120 Weff into 8 ohms.)
- I will study a stabilzed power supply for the heating and the amp in order to increase the ratio signal/noise (I believe that it is possible to win ~ 6 dB)

At the time being, i am working on a model in LTspice. It works not too bad, but it is not yet perfect.
 
Very nice. As a new owner of a MC240, I have thought about modifying the circuit (don't panic...I won't change the original) and then I read your post.

The Long Tail Pair could probably use a CCS and of course the power supply can be much improved. I was going to LTSpice the circuit and start from there. Do you have the VDV output transformer modelled in LTSpice?

Could you share the .asc Spice file here or via "conversation"? That would save me alot of work as the MC240 circuit is very similar.
 
I took some this morning and started on a LTSpice for the MC240. So far, the voltages are within 1% of what the SAMS Photofact shows them to be. However, without a transformer model I am at a loss to continue much further. If anyone needs the libraries referenced in the LTSpice file I can provide them or they can be downloaded.

You will have to remove the .txt from the file to allow it to be used. I had to add the .txt to "fool" the forum software to accept it as an upload.
 

Attachments

I couldn't leave this alone for now whilst thoughts of eating turkey later this week dance in my head. I think I have the transformer represented correctly using an excel utility I found on the web. The DC voltages are spot-on with respect to the SAMS schematic but as it stands I am not getting the output power I want by a factor of 100. I'm sure I'm missing something. I tried reversing the directions of the windings and that did not help.

I apologize that the picture is fuzzy, maybe because the forum software downsizes it? And I also wish LTSpice would let me truncate the dc voltages, I don't need it to the umpteenth decimal place!
 

Attachments

Last edited:
Yes, i can share the model I habe made in LTspice for the output transformer. It is probably improvable.

Here the content of the file :

******************
.SUBCKT VDV-1070-UC A1 B A2 K2 M2 M1 K1 S1 S2 FB1 FB2
* PLITRON VDV-1070-UC OUTPUT TRANSFORMER 4KOHM UC PRIMARY
* OL NUMBERS CORRESPOND TO TRANSFORMER SCHEMATIC.
.PARAM PRIML=1768 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM LRATIO={5/4000} ; INDUCTANCE RATIO: (5 OHMS)/(PRIMARY).
.PARAM QFCTR=1821000 ; Q-FACTOR: PRIMARY SHUNT L/LEAKAGE L.
LP1 A1 B {PRIML*.0625} ; PRIMARY
LP2 B A2 {PRIML*.0625}
LP3 K2 M2 {PRIML*.0625}
LP4 M1 K1 {PRIML*.0625}
CP1 A1 A2 .1745NF ; CAPACITANCE FROM SPECS SPREAD ON ALL TAPS
CP2 K2 M2 .087NF;
CP3 M1 K1 .087NF;
LP5 S1 S2 {PRIML*LRATIO} ; SPEAKER SECONDARY
LP6 FB1 FB2 {PRIML*LRATIO} ; FEEDBACK WINDING
K1 LP1 LP2 .9999989 ;
K2 LP1 LP3 1 ;unity coupled
K3 LP1 LP4 .9999989 ;
K4 LP1 LP5 .9999989 ;
K5 LP1 LP6 .9999989;
K6 LP2 LP3 .9999989 ;
K7 LP2 LP4 1 ;Unity coupled
K8 LP2 LP5 .9999989 ;
K9 LP2 LP6 .9999989 ;
K10 LP3 LP4 .9999989;
K11 LP3 LP5 .9999989 ;
K12 LP3 LP6 .9999989 ;
K13 LP4 LP5 .9999989 ;
K14 LP4 LP6 .9999989 ;
K15 LP5 LP6 .9999989;
.ENDS VDV-1070-UC

The schematic logic is explained below :

A1 : connected to anode 1
B: connected to + HT
A2 connected to anode 2
K2 : to cathode 2
M2 : ground
M1 : ground
K1 : cathode 1
S1 : Speaker 1
S2 : speaker 2
FB 1 : Feedback 1
FB 2 : Feedback 2

Here a screenshot :

upload_2015-11-24_23-38-14.png

I show the symbol. It is not very nice, but you just have to write VDV-1070-UC in the filed value for having the right calculation (obvioulsy, record the file above in a .txt and make the right LT spice directive : .INC yourfile.txt). I have not succeed for uploading the .sym. You have ti draw it again.

Some comments :

- I have spread the capacitance on the different taps.
- The winding resistance of the winding are not included in the model. You have to add it manually on the scheme (I have measured : A1-B : 19.6 ohm, B-A2 : 21.7 ohms, K2 M2 : 19,6 ohms and K1 M1 : 21,7 ohms)
- I have assumed a coupling ratio of 1 for the unity coupled windings and try to spread the inductance leakage on other windings.

Hope it helps you. It is a first step which is not perfectly matching with my measures (for the ringing)
 
I have looked at your scheme. I did not see the feedback loop. The values of your inductance seems very high on the output transformer.
For getting the max output you need ~ 0,7 V peak on the input.
You can also check the gain after each stage. Everything seems normal on your scheme.
 
Last edited:
The inductance did seem too high. The excel utility calculated that and I don't trust it. I did not have the feedback winding since I had no data. Thank you for the transformer model! I will look at it and see if that helps.

I need to mention that we all realize a model of this circuit does not eliminate the bread board and testing. However, it is a helpful tool and makes a draft circuit very quickly. It would be very nice to make this unity coupled transformer work with or into the classic Mac circuit with minimal changes.
 
Regarding the power issue, did you load the secondary in your simulation with a 5 ohm load? I believe that is the rated secondary load impedance for that transformer.

Dave
 
You must log in or register to reply here.
Back
Top Bottom