Transcendent Sound Forum
Latest Topics
 
 
 


Reply
  Author   Comment   Page 4 of 6      Prev   1   2   3   4   5   6   Next
Wolfgang
Reply with quote  #46 

Kelvin,

You are right, there is hardly any difference between full connection and the common ground version(see pics).  If it also sounds the same then indeed it would need only 3 switches.

Wolfgang

 

Achim,

I tried manual switching with SSRs and my 15 ohm C-Horns. No switching noise, clicks etc with double impedance settings. I could hear the difference between AT and no AT in-between switching intervals. But the ATs are not broken-in (maybe 20hrs). If I had used precise in/out switching at peaks only I am pretty sure that it wouldn’t have been audible anymore.

To describe the sound with the ATs is not easy. I don’t like it so much with my Lowther BLH because they sound more real without them. But with the KH/Lowther (see pic) it’s the opposite. With the ATs they sound so rich and indulging and there is no limit in dynamics and sound pressure. I don’t know if you have ever listened to the 250W Electrocompaniet amps with full range speakers . These amps were pretty popular in the 80/90ies. That’s the best comparison I can come up with at the moment.

Wolfgang

Attached Images
Click image for larger version - Name: Alesis Monitor double impedance.png, Views: 13, Size: 27.94 KB  Click image for larger version - Name: Alesis Monitor grey_wht-bl.png, Views: 18, Size: 27.66 KB  Click image for larger version - Name: KHPM6A.jpg, Views: 16, Size: 47.37 KB 

Wolfgang
Reply with quote  #47 
Kelvin,

I was wondering why you used a combination of 4093 and 4011/74S00 in your speaker protection circuit.  Or is there a reason (hysteresis?) why a Schmitt trigger would not work for driving the SSRs properly?

I am closing in on my little project and it still looks pretty good.

Wolfgang
Kelvin Tyler
Reply with quote  #48 
Wolfgang. The idea of the 4011 or 74LS00 is to form a once triggerable latch so that when the SSR is switched off it stays off. This latch is set to the correct state on switch-on via the RC delay at the second input. The Schmitt gates just help define clean trigger points.
Very interested to hear how your project finally turns out.
Kelvin


Achim
Reply with quote  #49 
Wolfgang,

the impedance plots in your last entry - is there really one with the AT engaged? Your first impedance plots showed very well the impedance flattening effect of the AT - strangely enough if I understand you correctly this was with the loudspeakers that did NOT profit from the ATs.

Anyway, it's very interesting to know that the switching in and out does not cause any "jumps" in response.

If we really crocked the workings of voltage and current in driving speakers and their drivers, then maybe we would understand what's going on.

Achim
Wolfgang
Reply with quote  #50 

Thanks, Kelvin. I will post the results as soon as I am there. I might have more questions at some point.



Achim,
The impedance curves in my last posting were from Alesis studio monitors (4ohm/88dB) not from the KH/Lowther. I just needed some speakers for a quick test of Kelvin's 3 switch solution and I grabbed the Alesis which were the closest speakers available in the room where I did the testing. I only added the picture of the KH/Lowther in order to give you a better feeling for how these speakers might sound as I simply couldn’t come up with a clear sound description at the moment. When I first posted the impedance curves they showed in the file name of the pics that they were from the Alesis monitors. Now the file names are gone. Sorry for the confusion.
The first impedance curve shows the Alesis with full AT connection for double impedance (yellow/blue-white-black). The second impedance curve shows the Alesis with Kelvin's proposal (common ground and white-grey) which is also double impedance . The difference between these two connections - using the lowest impedance reading as reference -is only 0,28 ohms and 1,5 degree in phase. So both curves show the Alesis monitor with double impedance, only the way of how the ATs had been connected was different.

To switch the impedance at peaks and allow the amps to use less current and more voltage  is the underlying idea of this project  and might show some new aspects of what we already know. Measuring current and voltage with real music signals with the SE 300B OTLs and speakers never really lined up with what I was actually hearing. They should have been able to play much louder according to the readings before they showed first signs of compression.  So there is some hope that even SE OTLs with a clearly defined/limited peak to peak voltage swing could be able to play quite a bit louder and more relaxed within a wider range close to their power limitations no matter how little the benefit of impedance switching might look on paper if we calculate it.

Wolfgang

Kelvin Tyler
Reply with quote  #51 
Just for the record, I am attaching a sketch of how I did the 8/16 ohm toggle switching with the Zero A-T and my TS Pinnacle 300B PP OTL. For this test I used the Fostex FE166En speakers to avoid any possible complication from cross-overs.
In this case there is very little change in SQ when the A-T is employed. Just a slight drop in perceived volume at constant input level and a very slight loss of open character to the sound. My Heresy speakers give a very similar result.
I did contemplate doing the switching automatically with the MOSFET switches toggled  with a low frequency multi-vibrator to facilitate the change over at various volumes, but,as I will not be using the A-Ts with the Pinnacle, I have not tried that at this time.
A-T switch.jpg 
Kelvin

Bruce Rozenblit
Reply with quote  #52 
Guys, the auto transformer is a step down device.  Any audio that runs through it will be reduced in amplitude by its turns ratio.  The only way to make it work is to increase the signal level going into the amp by the same amount of the reduction.  This scheme will function as a signal compressor.
Kelvin Tyler
Reply with quote  #53 
Agreed absolutely. The overall voltage gain of these amplifiers is virtually totally determined by the degree of global NFB. A figure of 4 or 5 is typical.
Kelvin
Achim
Reply with quote  #54 
Bruce, isn't the signal makeup needed against the step-down of the autoformer fully provided by the amp's negative feedback?
Bruce Rozenblit
Reply with quote  #55 
Nope.  That would only happen if the transformer was placed inside the feedback loop.  That is much more difficult to switch in and out because of the huge amount of reactance aded to the amp and would require completely different feedback compensation.  The only way to do that reliably would be to redesign the amp as an autotransformer output amp.  Then it isn't an OTL.       
Kelvin Tyler
Reply with quote  #56 
As ever I am very willing to be corrected, but I see things this way.
The 300B SE OTL amps whether AC or DC have been designed to provide a power output of 1W into a load of 8 ohms. They run in strict Class A with an total idle current of 0.5A. The maximum possible 'undistorted' rms power is then simply:
(0.5x0.5x8)/2 =1W. With a 4 ohm load there is no way to directly get 1W. The maximum power would then be 0.5W. To get 1W into a 4 ohm load a transformer of some kind with a root 2 turns ratio, can be used to reflect an 8 ohm impedance to the amplifier. An ideal transformer, whether auto or double wound, acts as a constant power device of course. The drive requirements in this case would be no different from driving an 8 ohm load directly.
Within certain limits, OTL amps can give more power with higher impedance loads, but, as Bruce points out, more input drive is required. One might think that it would then be a simple matter of getting 2W,say, by increasing the load to 16 ohms. This would require more drive and, in these amps, there is quite a low B+ voltage available to the 12AU7 driver tube. There may well be other limitations imposed by the cathode follower output stage. Grid current problems always have to be looked out for.
Personally, I find ten minutes spent with a scope and signal source looking at these and other amps very revealing.
Kelvin
Wolfgang
Reply with quote  #57 

My original idea was, simply put, that switching ATs at peaks would use the existing power of an OTL more efficiently in the volume range where we listen 80-90% of the time. Often classical music pieces have only some peaks that make it impossible to listen at a higher volume level which is on the other hand necessary for the quiet passages to sound realistic enough. Relatively speaking we would have the same experience like with a little more powerful amp (what I called “extra power” with quotation marks in the headline of this thread) if these peaks wouldn’t be compressed and distorted. Absolutely speaking and in line with the laws of physics we would still listen to the same 1W/8 ohm or 0,5 W/4 ohm OTL.

I tested this circuit today with a very rough input (resistor and cap/line level signal) and  a solid state amp connected to the Alesis monitors because I am still waiting for the VU meter kit. It works but the switching with this kind of input (neg. half wave of the music signal) is far away from being precise enough for precise triggering. Like this the difference in volume is slightly audible and I also got some minor switching noise transients which I didn’t get at all with manual switching in my main system with regular relays, 5 switches, the SE OTLs and 15 ohm Lowther.

At the moment I cannot say what could be the source for these transients. Maybe it comes from amp reactance (I used 6 and 12 ohm impedance option at the amp speaker terminal but with the same results), from using 3 different brands SSRs or simply because the AT needs to be switched with 5 switches and not 3? For the final signal input I will use the Velleman k4306 LED VU meter which provides 1dB steps. And I will go back to five switches, all from the same brand.

Attached Images
Click image for larger version - Name: e-POD1.png, Views: 14, Size: 58.10 KB 

Kelvin Tyler
Reply with quote  #58 
Wolfgang. A couple of points regarding your switching circuit.
I am not entirely sure just how you activated this circuit from your line amp R/C arrangement, but I can see that a high at the input gives normal OTL operation and vice-versa. I guess the idea with the attack time pot is to control the interval when a slowly(?) falling level at the input makes the first Schmitt change state. This then begins to charge the 1mfd capacitor. This is not recommended for CMOS chips in my view. They become very slow with capacitive loads. The output of the release gate eventually goes low and the switch to A-T mode is enacted. I am not certain as to the function of P3 here. When the input is released at the passing of the peak, the 1mfd cap discharges via the shunting pot. The charging problem then arises at each and every subsequent peak in the music. All this rather negates the point of setting an attack interval in the first place. I take it you are using 5V for Vdd on the chips. This circuit uses the CD4093 simply as inverters; a single Schmitt hex inverter package such as 4548 or 40106 would serve equally well.
I note that in most cases the nand gate inputs have been tied together. This generally changes the trigger points a bit.
The rough rule of thumb I always thought applied to companders was 'fast attack- slow release'.
Hope I have not got this totally wrong!
Kelvin
Wolfgang
Reply with quote  #59 

Kelvin,

You can activate the circuit with DC from the VU meter (levels that normally activate LEDs) or simply by feeding a line level signal through a resistor (for adjusting the input voltage) in series with a cap. But like  this it is not very precise switching, only can  be used for testing the general operation of the circuit with music.

Just for the sake of correct terminology and to avoid misunderstandings :  This circuit is not a compressor but uses some functions that are similar to a compressor. I use the words “attack time” and “release time” in order to describe specific functions of this circuit not as they are originally used to describe the function of a typical compressor.

The  trigger level for the first Schmitt trigger (“attack time”) should be adjusted with P1  to a level which is low enough so that only high enough DC voltage will be able to trigger from low to high. Setting this level applies to all selectable threshold levels the same. The threshold level equals the LEDs in the VU meter (-3dB…+4dB etc) and can be  selected by “threshold” settings.  Adjusting the trigger level low enough with P1 makes sure that very short DC peaks at any chosen threshold level will not activate the input because we don’t want too much switching action, only when the amps would start distorting at a long and high peak.

The “release time” pot P2 makes sure that the second trigger doesn’t  stop working while the peak still needs to run through the ATs. This time window is set by discharging a cap slower or faster and defining in this way how long the second gate stays high. This doesn’t need to be too fast.  A little slower is ok here.  How long or short this time needs to be set depends on the music but in most cases it needs also to be pretty short so that we activate the ATs only form start to end of peak signals (more or less).

P3 is necessary for adjusting the threshold for the final switching part more precisely in relation to the high state of the trigger for the “release time”. This has to be set only once so that as soon as the “release” trigger goes to high the SSRs start switching.

If the input is not triggered SSR2 /"OTL direct" is always ON as default.

Wolfgang

Wolfgang
Reply with quote  #60 
So much has been said already that I thought a visual would be helpful.
I uploaded a video clip that shows the working of the switches at fast settings. This is how it would look if the threshold is set relatively low (lots of triggering so that there is something to watch). For the final use with speakers  it would be set higher.
Triggering is not 100% accurate without the input from the VU meter. The green LED stands for the OTL direct, the red and yellow for AT IN/OUT.

Here is the link:

https://www.dropbox.com/s/uh46jsafyl4hk0n/fast%20settings.AVI?dl=0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.