Reply with quote #77
I guess you are looking for a fast up /slow down type of action similar to your 'release time' arrangement. I would do this before the final Schmitt drivers rather than slugging the input to the SSRs. It is best to present as snappy an action as possible to the SSRs. You can of course switch MOSFETs in nano-seconds, but that would require a bit of a redesign. Kelvin
Reply with quote #78
the switching speed of the Schmitt triggers and the SSRs is ok in my opinion. My goal is to keep the AT SSRs still open (delayed turn off) after the trigger has already gone from high to low so that there is no interrupt due to different turn ON times of the "OTL direct" SSRs and vice versa. You can see it works if you compare the scope readings from post #72 and #76. I am just mildly shocked that even 4ms more are not enough – if my calculation is right.
Interesting also in this context that the regular relays of my original test which didn't make any switching noise had an average of 12-15ms turn on/off time. The only way to increase delay time even more in order to eliminate even the tiniest residue of noise without getting the loudness change due to double ground connection of AT/OTL direct would be a switch for the AT ground (switch # 6!) which would operate without turn off delay while all the other SSRs would operate with delay. If this wouldn’t work and wouldn't make the circuit dead silent it means it cannot be done with SSRs.
Reply with quote #79
Browsing through the ZEROs manual earlier today, I noticed that the symmetrical 2:1 impedance connections do not in fact use the whole winding. In post No.73 I had wrongly assumed that they did, and so the comments I made there in this context were incorrect. The total number of turns used in the 5-switch and the 3-switch schemes are the same, or very nearly so.
I repeated manually switching from OTL to having the A-T connected with my Pinnacle amp,an 8 ohm speaker and with constant input level. In these tests there is a small, but distinct, drop in volume in the 3-switch setup, yet virtually no drop with the 5-switch arrangement. I do not really understand how this difference arises. Anyone have any ideas on this? There is more to this project than meets the eye (mine at least!). Kelvin
Reply with quote #80
In the 3 switch version we have parallel to the voice coil the AT making a ground connection. If we switch to OTL direct it’s only through the voice coil. In the 5 switch version it’s in both cases only one connection to ground through the voice coil/AT or voice coil OTL direct. This is the only obvious difference between the two versions. In the impedance readings we can find a little but distinct difference between 3 and 5 switch version as I have mentioned before. That looks very little (0,3 ohm, maybe more or less at different frequencies) but obviously it’s more noticeable than it appears considering only the tiny amount of impedance/phase change.
The whole current vs voltage drive of speakers is a discussion of its own. But that’s what we are doing, or at least it points in the same direction as this discussion IMO, if we switch impedance. The effect of this might be more complex than meets the eye because it includes loudspeaker parameters and the speaker design as well. Some go so far and even consider the room-speaker-air-volume interaction as part of the speaker parameters which will appear at the amp's output as reactive force if I have understood these studies right.
On the topic of switching I can finally conclude that at least with SSRs it’s always a compromise between more loudness difference and less switching noise and vice versa. The switching noise can be brought down to hardly noticeable levels (increase of delay time for the SSRs for switching OFF) but then loudness difference through double ground connection becomes more prominent which is even worse than the switching noise itself. Depending on the music style I can hear the little dull clicks more or less, sometimes they simply disappears in the music but this is not a solution for High End or if one aims for perfection.
Reply with quote #81
If I dare make a final point on impedance ratios with the Zero A-Ts. If we take the 2:1 and 4: 1 figures given in Fig.1 of the manual to be exact for the connections indicated, and we assume overall symmetry in the winding, then the impedance ratios for the 5-switch and 3-switch cases are not exactly the same. The respective ratios are 2:1 and 2.25: 1. This may well account for the small difference in Wolfgang's measurements which his elegant graphs show, but would hardly account for the respective volume drops. I wonder just how exact are the tappings?
I do feel that Wolfgang should have the last word (if there is one!) on this interesting project. Quite a bit of ground has been covered and I greatly admire his persistence in seeking a good outcome. I have learned a lot. Many questions answered, ideas justified, and so very, very close. Well done Wolfgang! I always thought that acoustic design and the like involved black arts. Kelvin
Reply with quote #82
Thank you for your kind words, Kelvin. Yes, it was a roller coaster ride but that’s anyway my experience with building electronic circuits. You never can be 100% sure that it works until it really works. But it’s always fun no matter how the outcome is.
I have a quote hanging over my work space where I do my little experiments which always cheers me up if things don’t go my way. It’s from T. Edison and everybody knows it: “I haven’t failed. I have just found 10,000 ways that won’t work.”
Reply with quote #83
I completely understand your outlook on this subject. What the amplifier is really seeing is a reflected impedance, just like it 'sees' the other, more common coil of the same, to which we often refer as the voice (or speech) coils of whatever dynamic drivers are in use with the amplifier in question. The crossovers of the older Klipsch speakers, such as the Klipshorn and La Scala, also imply an autoformer on the midrange driver for the same reason Wolfgang has shared -- where the auto former behaves as an attenuation device to match the output of the midrange horn with that of the tweeter and horn-loaded woofer -- and does so without the deleterious effects (subjectively) of an L-pad composed of either a single series resistor, or a series + shunt (to maintain original nominal impedance of the driver.
Reply with quote #84
Originally Posted by
erik Capt. Z I completely understand your outlook on this subject. What the amplifier is really seeing is a reflected impedance, just like it 'sees' the other, more common coil of the same, to which we often refer as the voice (or speech) coils of whatever dynamic drivers are in use with the amplifier in question. The crossovers of the older Klipsch speakers, such as the Klipshorn and La Scala, also employ an autoformer on the midrange driver for the same reason Wolfgang has shared -- where the auto former behaves as an attenuation device to match the output of the midrange horn with that of the tweeter and horn-loaded woofer -- and does so without the deleterious effects (subjectively) of an L-pad composed of either a single series resistor, or a series + shunt (to maintain original nominal impedance of the driver. *edited to correct spelling errors
Reply with quote #85
And there is -of course – another solution next to OTL direct with not enough power for certain speakers or dynamic peaks or autoformer coupling.
Find the CDF (critical damping factor) for your speaker and adjust the NFB of your OTL to match it.It worked with my C-horns in a most wonderful way.
Please read the full article if you are interested:
For the 15 0hm Lowther DX4 in my C-horns the ideal CDF is 2,5 ohm but adjusting the 300B SE OTL to 9 ohm sounds best with a lot more (acoustic?) power. That might be also one of the reasons why the MP with 15 ohm output impedance sounds so good with your 15ohm Lowther Medallions because they simply need a lower DF/higher output impedance than the original OTL with R15=2k (0,8 ohm?).
Here is the measured CDF for the C-horns at 2,5 ohm and 9 ohm.
Reply with quote #86
Wolfgang - thanks a lot, great example of the difference between theoretically best and perceptively best...
As there are in a source-amplification-speaker-room-body-ears-mood system so many damping and resonance factors, the ability to tuning in by ear is of so much value. In my opinion every hifi device should have one rotary tuning knob. :-)
Reply with quote #87
all of my SE amps have such a pot in the meantime. Thanks for pointing this out to me some time ago (CDF and other speaker related stuff)! And it even gets better. Encouraged by the good sounding results with my 300B DC OTLs I embarked on a project which Ray brought to my attention some months ago: A SE OTL with 6SN7/6C33C (1 output tube only!) based on some ideas by J. Broskie. I optimized it for 15 ohm drivers (9ohm output impedance). It's truly dead quiet (great PSRR in the prestage with J.B.'s design) like my Mark Levinsons which I once owned. The music manifests out of nowhere on a pitch dark background which one normally only finds with solid state amps. It's sounds like 100W per channel while it is only 1,5W into 15 ohms.Bass is outstanding, lower mids incredible, treble like the best of the best.Not the slightest hiss/subtle noise around single notes, just space. If you ever need an amp for around 15 ohm speakers...this is it. And it has a pot for adjusting the NFB (see pics)!
Reply with quote #88
Wolfgang - you have an admirable ability to take things to the end and then further.
Reply with quote #89
Yes, Achim, Wolfgang's experiments often uncover some very interesting results. I very much like the the look of his 6C33C amp and, indeed, I have been considering constructing a one-channel version of this for some time.
My experimental approach is generally based on using external power supplies and 'modular' units. By way of example I attach a couple of photos showing a cascaded 6SN7 driver unit, and a 2x EL509/Lm317T output module which can give ~1watt into 8ohms if each tube is set to idle at ~250mA and bridged. NFB is adjustable by varying the resistor/cap components connecting the output to the cathode of the first 6SN7 triode. As Wolfgang suggests a pot is very handy here. The sound of this setup, when optimized , is very pleasant. I have ordered the bits and bobs to make the 6C33C module,and will use my external power supplies and the 6SN7 driver in a similar fashion to what I have described above. Personally, I find this approach to be quite economical for my many experiments. Kelvin
Reply with quote #90
Hi Kelvin - you make me envious!
I have often thought of having something like this, but never got round to putting things together. And probably never will, either... The Wago terminal connectors would come in handy for quick connection and disconnection within the circuit for such modular approach: https://eshop.wago.com/JPBC/0_5StartPage.jsp?TopNavi=0_6TopNavi.jsp&Zone=2&Hauptframe=%2FJPBC%2FCommonPageHandler.jsp&activatedPage=SEARCHPAGE So you can quickly try different parts without soldering. I replaced all the terminal blocks in my Pinnacles with sets of Wago connectors ;-) My estimation is that I needed less time to put together the two Pinnacle mono blocks than it took to solder the Masterpiece. Cheers, Achim