Transcendent Sound Forum
Latest Topics

  Author   Comment   Page 3 of 7      Prev   1   2   3   4   5   6   Next   »
Reply with quote  #31 
Kelvin & Achim,

I cannot say much about SQ at the moment as the A-Ts need to break-in a little more. What I can say is that there is absolutely no switching noise, amp reaction, or anything that would point in the direction of what we thought that could happen.

For a first test I used a simple toggle switch to switch back and forth between A-T and OTL direct. The only thing that was clearly audible was that with 1:2 AT it was about 2-3dB louder and there was a tiny bit more electrical noise audible right at the speaker. There were some more audible things but it's too early to say something and it might disappear after some breaking-in.

The transition between switching was so seamless that I wasn't sure even if my switch was working. So I checked the phase/impedance curves (see attachments). It really worked and the transitions are absolutely seamless no matter at which volume.

Soon I will have more details.


Attached Images
Click image for larger version - Name: KH_PM6A sine AT switched.png, Views: 21, Size: 348.61 KB  Click image for larger version - Name: KH_PM6A sine AT.png, Views: 20, Size: 353.71 KB  Click image for larger version - Name: KH_PM6A sine.png, Views: 19, Size: 316.31 KB 

Reply with quote  #32 
Maybe it worked so well because I only used SW1 &3 but not SW2. The A-T was during switching always connected to the speaker.
Kelvin Tyler
Reply with quote  #33 
Your impedance/phase sweeps show the expected changes when switching the A-T in and out. So far your observations are consistent with some of the comments on the Anticables web site. I am still waiting for my A-T to arrive; stuck in UK customs I imagine as usual.
Will renew my considerations re my Masterpiece/resistor observations. The resistor loads I used (62 and 124 ohms) were much lighter than an 8/16 ohm speaker. The output impedance of the MP is about 15 ohms.
I guess your next move might be to see if the peak transient effects you mentioned in a previous post are still noticeable, at your normal listening level, and with the A-T switched in. You could try this now with your toggle switch setup.
Good luck.
Reply with quote  #34 
The impedance curve (switched) was already hinting at the situation when using only 2 instead of 3 switches as it is about 2 ohm higher than the original reading without the AT. I would call this a "hybrid" solution. It's not pure OTL because part of the signal goes obviously still through the AT via ground when using only 2 switches but it's an easier load already compared to pure OTL without the full sound effect of the AT. That's why the loudness difference was only 2-3dB, less than I had expected.
But both versions sound different, both really good.
It needs all 3 switches in order to get pure OTL on one side and- after switching -OTL through AT on the other. The loudness difference is then about 6dB and the impedance curves are correct again but there is still absolutely no problem with switching. The presence of the AT is audible because of some tiny amount of more noise/hiss.
Reply with quote  #35 
Great results! Looking forward to hearing your description of the A-T "sound effect"...
Kelvin Tyler
Reply with quote  #36 
Yes indeed, excellent results.[thumb]
A couple of questions, just to satisfy 'the insane and feeble wanderings of my fading brain' ( sorry Shelley).
I take it the 1:2 A-T ratio you refer to in post no.31 is the impedance ratio i.e. a winding ratio of root 2, ~1.4. From your original SW1/SW3 impedance sweeps I see there is an approximate doubling of the impedance when the A-T is switched into action. It interests me that you say shunting the speaker with the otherwise disconnected A-T increases the impedance by about 2 ohms. Is this because there are caps in your cross-over? How is the impedance measured exactly?
I note the 6dB increase in loudness. Is this with the same drive level? If so, would this not be quite noticeable with a sustained peak that just exceeded the threshold?
I imagine one might have to be constantly changing the threshold according to volume setting and music type. Personally, I have problems deciding on the appropriate settings when adjusting a Schiit Loki mini tone control! I presume you will be using your excellent PC/Soekris/Triode feed with both digital and NFB volume adjusters. You certainly revel in adjustable parameters.
All very interesting. Look forward to hearing your further discoveries, and, indeed if you choose to go for the full switching scheme, or, as Achim suggested, the choice may be mooted.
Good work.
Reply with quote  #37 


Very funny to quote Shelley in this context!

I am always referring to impedance ratio, not winding ratio. I’d better call it multiplication factor 2 which simply doubles the impedance.

I use a 1st order x-over with these speakers..just caps and resistors for smoothing.

The little program  comes with ARTA, it’s called “Limp”. Here is the link: You need a computer with full duplex sound card and a little breakout box.

My dB levels were based on guessing, sorry if that wasn’t clear. I also used different kind of music with different dynamics, bass  etc. This was not a real test in absolute terms. It’s also hard to say anything definitive in absolute terms as the ATs influence the sound across the whole frequency spectrum. When I play music with the ATs it always gives me  the impression of “louder” than it might be in absolute  dBs simply  because the bass sounds more powerful, treble more brilliant, midrange more 3D than direct OTL with hard to drive loads.  I didn’t have this impression at all when I used for one channel my C-horns with 15 ohm Lowther and on the other one the speaker with AT. If I use it without AT it falls back in volume and resolution like playing from farther away.  I don’t think the switching would be very noticeable if it happens only around peaks. But it might be more audible with certain kind of  music.

The accuracy of the switching has nothing to do with my DAC volume settings or line stage but depends only  on the precise settings of threshold, attack and release time. It's pretty simple: if you once have set any threshold (like -2dB for example) which would reflect the max volume which your OTL-speaker combo can handle without problems then everything that hits or exceeds this mark will go through the AT while below this mark it's pure OTL. Once set like this for max level at -2dB the level meter would simply reflect your listening volume settings (most of the time below -2dB) without the need to change anything. The whole idea is that the AT will only kick in when we exceed the threshold which is not defined by the music but by your amp-speaker and the highest possible volume for it in combination with your actual listening level. The louder you listen the more often peak levels  will go through the AT.

I can already say that I will  use the ATs without any switching with my KH and KH/Lowther. My 15 ohm Lowther doesn’t need any help or boost, and the only candidate for switching would be my 8 ohm Lowther Medallions where I originally got the feeling that they could perform better at high volumes with a more powerful amp.


Kelvin Tyler
Reply with quote  #38 
Hi Wolfgang,
Thanks for a very clear description of the threshold setting philosophy. Yes indeed it has to be set according to the capabilities of the power amp/speaker combo.
Simple matter then to back off the volume a shade if the switch over became noticeable on any sustained peaks just exceeding the threshold.
Kelvin Tyler
Reply with quote  #39 
This is proving a good example of the value of hands-on experimentation.
I must confess that I was somewhat sceptical about this project for some time.
I realized that an A-T could 'improve' the performance of an OTL amp, but have to plead guilty to have given a 'bum steer', at least on one occasion. This is quite apart from the minor 'misunderstandings' that are inevitable on any forum.
The gain of a cathode follower (CF) can be expressed, to a good approximation, by: gm.Rk/(1+gm.Rk), where gm is the tube mutual conductance and Rk is the load resistance. A glance at this indicates that the gain approaches unity as Rk is increased, but that for small values of Rk, such as a speaker's 4 or 8 ohms, the gain is closer to gm.Rk. I wrongly referred to the 'high resistance' situation when I said the power in the load is halved on doubling the resistance. With low resistance, heavy loads as here, the power output is approximately doubled.
Of course, gm is hardly constant over the whole operational range, and the power output is ultimately limited by the current capability of the tubes. You will not get anything like double the power, i.e. 2 watts, near the tube limit. For four 300B tubes in parallel the effective nominal gm is about 22mA/V, which with an 8 ohm load gives a gain of about 0.15. Apologies to those on the forum to whom this is all 'coals to Newcastle'.
There is still quite a distance to run with this project. With Bruce setting such a high performance level with his designs, this may turn out to be another example of the effort required to gain just a bit of enhancement. Hardly a commercial option.
Wolfgang is certainly an 'ideas man'.[smile]
Reply with quote  #40 

I would rather describe this “project” as an opportunity to look at the same thing ( OTL with difficult speaker loads) from different angles.  And the reality shows that this is an ongoing topic on this forum: Can this OTL drive my speakers? I am only listening at moderate volume levels, can this OTL drive my speaker which should normally not be used together with an OTL?

Kelvin describes it mainly from the amp’s perspective (please correct me if I am wrong!). I tried to look rather  from the speaker’s perspective. The difference between these positions is that from the amp’s perspective there is only one truth (or at least it seems like that), from the speakers perspective which includes the amp - but not as something static but as an always changing component in relation to the speaker - it can look a little different and of course a lot more complex. Doesn’t have to in every case but can in certain cases.

I think the “speaker’s approach” is also the position of the company that sells the ATs: “ The Zeros are not an amplifier fix, it’s a speakers fix.”

Let’s summarize a few things for more clarity: If we double the speakers impedance we change its sensitivity but not its efficiency. This means the speaker will draw less current but instead use more voltage for the same watts. But isn’t current delivery into low impedances the problem with OTLs?

If I hook up my 15 ohm Lowther with the ATs (30ohms then) it shows exactly these results. They play a tiny bit less loud and of course there is in addition the audible effect of the AT. There is absolutely no benefit with this combination. Pure OTL without AT is always better under all circumstances (music style, volume etc).

If I hook up my original  KHs I also get a tiny bit less gain (didn’t expect that) but at very loud levels and at certain frequencies the clipping doesn’t happen as it becomes audible in the OTL direct connection. In this case I am not sure yet after some listening with different music styles what’s better because the OTL direct connection can play louder and it might be enough in most situations but they also can clip more easily. The KH sound a little more “saturated”with the ATs  and it’s  a matter of taste what one would prefer. I am sure borderline  cases like these would  also show different results with different OTLs!

The only dramatic improvement so far was with my KH/Lowther combination  (very simple x-over compared to the original KH) which can play a lot louder and with more satisfying overall sound with the ATs than with OTL direct. This is a clear case of 100% improvement at least for my taste and at least with these amps.

In the manual that comes with the ATs there are cases listed when an OTL can play quite a bit louder with the ATs. I just quote it but leave it up  to the experts to decide if that can be true: “Actually some systems will have the ability to play louder if the amplifier being used has a maximum power band curve that peaks around 16ohms. Some OTL examples are the…Atma-Sphere M60.., and the 25W TS T8 that should do 40W into 16 ohms.”

Kelvin Tyler
Reply with quote  #41 
My 'Zeros' A-Ts have finally arrived! As expected, they give no advantage when I tried them with various impedance settings with my Pinnacles driving my Heresey III speakers. They are pretty transparent though.
My 300B DC SE  is out on 'permanent' loan so I cannot comment on using that right now. As an amusing aside, these A-Ts do enable my Masterpiece to drive the above speakers to just listenable volumes - OK for late night use!
I have tried putting LCA715 MOSFET switches in the speaker feeds from my Pinnacles. No noticeable loss of SQ and switch cleanly. As I will not be implementing the switching scheme that has been the main subject of this thread, I may just try a couple of LCA715s in an additional speaker protection role. This is hardly necessary for Bruce's OTL amps and is certainly not an original idea. However, just for amusement, I attach a putative sketch of what I had in mind. The chip types are simply what I have to hand and the final switch activation would possibly need a small transistor connected to the other output of the latch to give reliable action.
speaker protect.jpg 

Reply with quote  #42 

No OTL needs ATs as long as the speakers are a good match (like Heresy and Pinnacles).  The basic idea of this thread was to address those cases (probably the majority of OTL users) when speakers cannot be driven properly by the OTLs at higher volumes or at peak signals. Often we only find out in A/B comparison that the OTLs don’t deliver full dynamics because of limitations in connection with the speakers we use. I feel somehow that my original idea hasn’t been understood fully. For sure it didn’t spark over but that’s ok as long as everybody is happy with one’s setup.

The original function of a compressor is to cut off peaks above a threshold so that the average signal level can be set higher without distortions. To use a compressor as control logic for switching ATs would do the opposite. It would make it possible for the OTLs to drive the speakers during peaks without distortion (not cutting the peak signals off!) and as a consequence we could listen at louder average levels. An SE amp will always deliver the same max power of course but the 10-20% peaks in a song wouldn’t define any longer the upper limit of the amp-speaker combo but the “louder” average level would be the defining element. That would sound in the end like a more powerful amp. Very simple and practical. PP OTLs might be able to deliver even  more power into the speakers with ATs  (switched or unswitched) but that would depend on certain factors which would need to be checked for each case.

I also have tested the LCA715 SSRs and couldn’t hear any audible effect or switching noise. However, it needs 5 switches (not 3) because the AT needs to be completely disconnected from the speakers for best SQ with OTL direct.  

I was side tracked by something which turned out to be very interesting. It actually goes a little in the direction of what Kelvin tried with the MP and AT.  I thought I should finally find out how a 300B DC might sound with luxuriously less NFB (12dB in this case) as the lower output impedance could be matched by the ATs easily. With the 15 ohm Lowther the ATs weren’t necessary even for high volume levels but the sound really improved a lot in this case. The max power and amplification of the amps is still the same but as the input signal is less reduced by the less applied NFB the amps responds surprisingly different.

Reply with quote  #43 
Of course it's lower DF  or higher output impedance with less NFB.
Kelvin Tyler
Reply with quote  #44 
Wolfgang. IMHO, I do not think your original concept and object has not been fully understood. You were pretty clear in your first post as to what you wanted to achieve. Personally, I have found this discussion to be quite stimulating, but, for me at least, I find the subjects of peak limiting compression, volume compression and expansion, audio AVC, and the like, far from being simple. Some of these techniques are widely employed of course in broadcasting, record cutting, etc, not always with favorable results.
With regard to the need for 5 MOSFET switches to enable your scheme. I do not see the requirement to use the Zero A-T symmetrically with the 300B DC SE amp. You would get just over the 2:1 impedance ratio if you used just 3 connections rather than 4, ie use black as common ground, Gray from the amp, and White to the speaker. You could then just use the 3 switches as in my original sketch. I note that the 3x and 1.4x examples in the manual also 'break symmetry'.
Looking at Fig.1 in the Zero manual, the exact impedance multiplication factors given are not strictly self consistent with a symmetrically wound A-T, though the error is hardly significant in the context.
Reply with quote  #45 
To my mind, the interaction of output impedance, resulting so-called "damping factor", connected speakers and their impedance, especially its variation across the bandwidth as determined by the electrical parameters of the drivers(s) as well as the cabinet - and perceived sound quality is far from being easily defined and controlled. High DF = better sound is not a valid formula. High load impedance = less distortion and better amp output power, on the other hand, is certainly a valid formula.

I would like to hear from Wolfgang and Kelvin about their sound impressions with the ATs inline. And from Wolfgang I would also like to hear what happens at peaks when the ATs are dynamically switched in.
Previous Topic | Next Topic

Quick Navigation:

Easily create a Forum Website with Website Toolbox.