[zybar]

Really?  I find Ralphs technical comments interesting from what I remember.

Ralph only posted once in that thread (at least that's all I see) and it wasn't an overly technical post, it was about being compared to Futterman designs.

Am I missing something?   

George

[jtwrace]

Ralph only posted once in that thread (at least that's all I see) and it wasn't an overly technical post, it was about being compared to Futterman designs.

Am I missing something?   

George

I guess he removed a BUNCH of his posts then. 

[Clio09]

roscoeiii, the T-16 for the money, especially used which is how I acquired it, is quite a bargain. It certainly has a number of the qualities an OTL is known for. If you have speakers that can work well with a 1 ohm output impedance (think solid state amps) then I have no doubt this amplifier can provide quite high levels of satisfaction. With my speakers it sounded slightly anemic versus the Atma-Sphere. Yes it is half the power but even when I pulled a couple tubes from each channel of the S-30, I found the same results (also, I use the jumpers in my S-30 so the gain of the amp is very low and sensitivity is close to 6V). Biggest difference was the bass. Before comparing the T-16 to the S-30 I was always a bit unsatisfied with the bass of the T-16. After comparing it to the S-30 it was evident that the bass of the S-30 was not only more extended, but fuller and more natural sounding as well. Again, I run my S-30 without feedback and I also have the beefed up power supply.

FWIW I use the S-30 with a Resolution Audio Cantata and Otari MX-5050 BII tape deck. Both are connected directly to the balanced inputs (I swap) cables when I want to switch sources) and drive the S-30 via their internal volume controls.

[roscoeiii]

Hi Clio09,

Thanks for the update. Yes, I really like the bass on the S-30. Is the conversion to zero feedback a mod that Ralph does? There's a pretty low amount of negative feedback as is, IIRC. And one of these days I will scrape up enough cash to get the upgrade to 3.1 status (I have a 2). And then maybe it will be time for the power supply upgrade, or the caps, or...

[Roger A. Modjeski]

Roger,

There seem to be three popular methods of OTL. Futterman, Circlotron, and Rozenblit. Could you please comment on them ( for educating us) and the popular tubes used? There probably are only about a dozen people in the country that understand OTL amps.  Thank you.

 steve

Other than a few very low power single ended designs, OTLs are push-pull and usually not class A though they do, as all push-pull amps, have a class A region. The class A region may be just a few watts on a 100 watt OTL. More about that later. A single ended OTL is hampered in that the output current will be only twice the standing current. As the standing current is likely to be limited to 0.2 amps per tube the peak output power is that squared times the 8 ohm load resulting in 0.32 watts per tube. More tubes, more power. As you double the number of tubes you quadruple the power, hence it is tempting to have a lot of tubes.

In all push-pull amps there is a tube or bank of tubes for the positive half cycle and the same for the negative. The Futterman amps employ a single power supply around 300 Volts with the negative end on ground. He also employs a large output coupling capacitor (1000 uF) to couple the 150 VDC at the output point to the speaker.

In the Futterman the output tubes are arranged as a "totem pole" one above the other with 150 V at their junction. The top tube can pull up to 300 volts and the bottom can pull down to ground. The amount of current is about 1.5 amps per tube. The tubes idle at about 100-200 mA per tube resulting in 15-30 watts dissipation per tube. To get more power just put more tubes in parallel. I used 4 in parallel per bank in the SA-4.

In the SA-4 and the designs I am working on now I use a split supply of plus 150 and minus 150 V (referenced to ground) and direct couple to the speaker thus eliminating the output capacitor making it a DC amplifier. To keep it centered (zero offset) I use a servo and a protection circuit should a tube short or large DC appear at the output.

The Atmasphere amps are a Circlotron configuration which is essentially a bridge with two floating power supplies of about 135 Vdc each neither referenced to ground. The output has a virtual ground at center of the speaker (if it had one). The original Electrovoice amp had this ground on the center tap of it's output transformer. The EV amps had an output transformer which Ralph eliminated by using a great number of tubes in parallel. I have worked on a MA-1 which has 12 output tubes per channel. Although Ralph feels tube matching is not important, I did match the output tubes so that they shared current equally. Otherwise some ran hot and others cool, shortening the life of the hot ones.

I note that the Atmasphere S-30 has an odd number of output tubes per channel. I have not seen a schematic of this amp and would be interested in looking at one. I suspect the odd tube may be the driver for the other parallel output tubes and Ralph may be going into positive grid voltages to get more power. Any comments?

As to Rozenblit's cicuits, the early ones I saw in Audio Amateur and Glass audio were of totem pole (Futterman) design but he was trying to improve on the driver. There were several mistakes in that article.  I don't know what he is doing now. Again, a schematic or perhaps the patent number would help.

[roscoeiii]

Roger thank you for your contributions here as well as starting this quite informative thread.

[roscoeiii]

And Roger,  I notice you added an autoformer question to the poll. Maybe you don't want to bias these results, but eventually, I'd like to hear your thoughts on autoformers with OTLs.

[zybar]

The Atmasphere amps are a Circlotron configuration which is essentially a bridge with two floating power supplies of about 135 Vdc each neither referenced to ground. The output has a virtual ground at center of the speaker (if it had one). The original Electrovoice amp had this ground on the center tap of it's output transformer. The EV amps had an output transformer which Ralph eliminated by using a great number of tubes in parallel. I have worked on a MA-1 which has 12 output tubes per channel. Although Ralph feels tube matching is not important, I did match the output tubes so that they shared current equally. Otherwise some ran hot and others cool, shortening the life of the hot ones.

Roger,

Matching output tubes can definitely make a noticeable difference.  A friend did this for me and it was easy to hear the improvement.

BTW, the newer MA-1 amps have 14 output tubes per amp and 5 6SN7 tubes.

Thanks for allowing us to talk about competing products on your Circle.  That is a classy thing to do and shows your level of maturity, as well as confidence in your products.

George

[OzarkTom]

No bass? You should have been to the CES show back in the 80's when crazy Harvey Rosenberg, rip, brought in The Quad ESL-63's with the NYAL OTL 1's. Every person that walked into his room looked for the subwoofers that he had none of. It was the highlight of the show. No one knew Quads could sound like that.

Roger, do you remember when Harvey pulled that off?

[zybar]

And Roger,  I notice you added an autoformer question to the poll. Maybe you don't want to bias these results, but eventually, I'd like to hear your thoughts on autoformers with OTLs.

I use Zeros with my MA-1's and Vandy 5A's and the system sounds better when they are in the chain. Depending on your speakers, you might think otherwise. 

When I tried Zeros with M-60's and single driver, higher impedance speakers, I felt that they not only didn't make things better, they made things worse.  By worse, I mean that you could clearly hear that the Zeros had been added to the equipment chain and were negatively impacting the sound. 

George

[roscoeiii]

Thanks for allowing us to talk about competing products on your Circle.  That is a classy thing to do and shows your level of maturity, as well as confidence in your products.

George

+1

Look forward to seeing what OTL products you decide to produce.

[poseidonsvoice]

As to Rozenblit's cicuits, the early ones I saw in Audio Amateur and Glass audio were of totem pole (Futterman) design but he was trying to improve on the driver. There were several mistakes in that article.  I don't know what he is doing now. Again, a schematic or perhaps the patent number would help.

A simple google search for me showed his patent clearly. The patent number is: 5604461. The paper he presented to receive his patent is right here.

What he is doing now, is a significant departure from his older circuits in Glass Audio. The patent eventually became what was in his T8 amplifier. His Beast is his best application of the aforementioned patent, with some improvements, and more powerful as well.

Best,
Anand.

[steve f]

Thanks Roger, this topic is both informative and unique because most designers have an agenda of their own which does not include input from the circle. 

At one time, I would have been against using an auto-former, but considering your past clever transformer applications, and the potential to reduce the number of output tubes, I can only say go for it.

Steve

[opnly bafld]

No bass? You should have been to the CES show back in the 80's when crazy Harvey Rosenberg, rip, brought in The Quad ESL-63's with the NYAL OTL 1's. Every person that walked into his room looked for the subwoofers that he had none of. It was the highlight of the show. No one knew Quads could sound like that.

Roger, do you remember when Harvey pulled that off?

This thread might be of interest:
http://www.whatsbestforum.com/showthread.php?2025-Constant-power-tubes-vs.-solid-state-amps-and-wild-impedance-speakers&highlight=quad

Lin

[gnnett]

Roger
I think the S30 gets away with odd number of tubes because the tubes are 6AS7, which are dual triodes, so 5 tubes per channel equates to 10 triodes per channel.

Regards

Gnnett

[Roger A. Modjeski]

Roger
I think the S30 gets away with odd number of tubes because the tubes are 6AS7, which are dual triodes, so 5 tubes per channel equates to 10 triodes per channel.

Regards

Gnnett

I considered that which means the fifth tube would have to be split, one triode pull up and one pull down in the same envelope. Seems like a lot of trouble for one extra triode. I am wondering if the fifth tube is a driver, such as a cathode follower to the outputs. Anyone care to check into this?

[Roger A. Modjeski]

Thanks Roger, this topic is both informative and unique because most designers have an agenda of their own which does not include input from the circle. 

At one time, I would have been against using an auto-former, but considering your past clever transformer applications, and the potential to reduce the number of output tubes, I can only say go for it.

Steve

The greatest problem in making transformers for conventional push-pull amplifiers is the coupling of the primary sections to each other and to the secondary. I have spent years developing transformers that do this well and the ones in the RM-200 MK II are the best so far.

With an OTL topology, the transformer can become an autoformer and there is but one primary winding verses two in a conventional amplifier. The coupling problem goes away. Now the transformer can be outside the feedback loop making the internal feedback far more stable.

I have been experimenting with a totem-pole output stage with only one pair of tubes. The driver is also very simple, DC coupled and balanced drive with only two stages. This is far fewer than any existing OTL amp that I have seen. With a single pair of output tubes I can get 60 watts into 600 ohms with a good class A range of about 20 watts. Without the autoformer this same amp is a measly 6 watts.

Here's the Transcendent patent. It is far from simple.  http://www.audiodesignguide.com/otl/5604461_tif_1_s0_35_r0.gif

[Clio09]

roscoeiii,  the S-30 has 2 dB of negative feedback in stock form. Ralph, being very familiar with my speakers suggested I disable the NFB circuit and see if I liked the results. So he walked me through the process on the phone which was quite simple. I did like the results so I left the NFB circuit disabled. I also have the 3.1 upgrade.

[Clio09]

I hope I'm not opening up a can of worms, but no one has mentioned the Berning designs when it comes to OTL. I know Ralph and David have debated whether Berning designs are really OTL, but I think very highly of them nonetheless.

Any thoughts?

As for NFB in an OTL circuit. Is it really necessary?

[Roger A. Modjeski]

I hope I'm not opening up a can of worms, but no one has mentioned the Berning designs when it comes to OTL. I know Ralph and David have debated whether Berning designs are really OTL, but I think very highly of them nonetheless.

Any thoughts?

As for NFB in an OTL circuit. Is it really necessary

First I would like readers to temporarily drop all the notions they might have about feedback.

Here are the problems of amplifier design as I see them.

How does one achieve low output impedance which is essential for most speakers to sound good because they were very likely voiced on an amplifier with low output impedance?

The output impedance of the S-30 with 2 dB of feedback is 7 ohms. Without the feedback it would be about 20% higher making it about 8.4 ohms. I would venture to guess that any sonic difference you heard by disconnecting the feedback was due to the higher output impedance.

I do not care to make amplifiers with such a high output impedance and I generally shoot for an output impedance that is 1/10 of the load impedance. This means that the output voltage will increase only 1 dB if the load goes to infinity. The 7-8 ohm output impedance of the Atmasphere allows the output to rise 6 dB when the load goes high. This is a very audible difference as some loads go very high, usually in the bottom end.

Since the NYAL OTL/Quad 63 combination has been mentioned I have observed the following. A circlotron OTL was brought to my home and connected to my Quad 63's and there was incredibly large but not so good bass. At 50 Hz the Quads have a very high impedance and at high frequencies a very low impedance. I connected an AC voltmeter directly to the speaker terminals and observed an 8 dB rise at 50 Hz. This of course is very noticeable. This is not what Quad had in mind when they voiced the speaker with their very low impedance transistor amps. The top end was also rolled off about 6 dB due to the low impedance there. Yes, feedback is necessary and not a bad thing in a loop that does not include the output transformer.

Feedback also reduces distortion but since that comes down with output impedance I am more interested in setting the output impedance. Generally when I get this to my bogey value the distortion is less than 1% at full power and 0.1% at average listening levels.

As to the Berning design it is mostly solid state and looks more like a switching amp. It is not the kind of OTL we are discussing here.