| Search | Login/Register
   Home » Melquiades Amplifier » Planning my DSET (187 posts, 10 pages)
  Print Thread | 1st Post |  
Page 8 of 8 (187 items) Select Pages:  « First ... « 4 5 6 7 8
08-22-2019 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 176
Post ID: 25554
Reply to: 25553
AVC experience
Romy, have you had any experience with Autoformer VC's?  From what I can tell they could work quite well at the output of a preamp driving the DSET's, mainly because they lower the output impedance of the preamp as they attenuate rather than staying high like a shunt switched attenuator.  It seems that in this particular situation that an AVC is the only real option apart from a 350R or lower shunt switched attenuator, which I do not find realistic.

Otherwise, I think that the VC will have to go at the input of the preamp and there are plenty of types that are suitable there, but I have a tendency to want to attenuate the large signal at the back end of the pre rather than at the smaller signal at the input.
08-22-2019 Post does not mapped to Knowledge Tree
Romy the Cat


Boston, MA
Posts 10,160
Joined on 05-28-2004
Post #: 177
Post ID: 25555
Reply to: 25553
Read sun is rising...
There are some commercial preamps that have sub 10R output: Karana, Quicksilver, Rogueaudio and many others. Some of them will ship the units to you to try. It might be good to experiment with them. To have you local guy who will work with you is a good direction.  I have no idea how the Muses chip sound, of how the PS Audio’s Stellar Gain Cells sound. It all needs to be tested. In most of the cases the claims the preamps makes make are not accurate, not because they are bad people who spread bad claims be because in most of the cases they do not know what results are expected buy you. It however might be the case where your expectations do not exceed the level of the expectations under which the preamp was designed/tested by maker. So, it is always the experimentations unless you know the preamp provider personally and know what level of reference point he supports.
With most of the none-discreet SS devises nowadays the gains is not problem. They use some kind of integrated circuits that run a lot of feedback. Levering the feedback, changing one resistor will yield more gain. I would like to sound as some kind of purist and to make the claims that integrated circuits are crap and global feedback is crap. The reality might not support it and I feel that everything is on a table. My tuner and my DAC have a few dollar worth op-amps in output stage and you do not see me complaining. 
 
I still do not know what is your problem with Placette is, it is a mystery to me. I know that Placette nowadays has a new version, very different output devises then it use to be and it has gain. I have both and new (believe me or not) in the same box. I feel that there is a difference in sound between the old and new. The new one is very transparent in term of Sounds but in my view it has some aria of concerns as well. 
 
You might go an alternative route. Why you are in progress to fine a good preamps you can get bad one that meet your specification and use if for a while. That what I would do. Find any cheap commercial preamp with a few Ohms output impedance.  Most likely it will be some mass market Japanese things from 80-90s, like Denon PRA-1200 or the similar. The eBay should be your best friend and the key is to get one under $100. As you get the unit you will bypass all bad input switches, tone controls, filters and the rest crap they build in and you only the preamp section. It will not sound good (most likely) but it will let you play you playback and to see what else is out there.


"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche
08-22-2019 Post does not mapped to Knowledge Tree
Romy the Cat


Boston, MA
Posts 10,160
Joined on 05-28-2004
Post #: 178
Post ID: 25556
Reply to: 25554
I do not know.
 anthony wrote:
Romy, have you had any experience with Autoformer VC's?  From what I can tell they could work quite well at the output of a preamp driving the DSET's, mainly because they lower the output impedance of the preamp as they attenuate rather than staying high like a shunt switched attenuator.  It seems that in this particular situation that an AVC is the only real option apart from a 350R or lower shunt switched attenuator, which I do not find realistic.

Otherwise, I think that the VC will have to go at the input of the preamp and there are plenty of types that are suitable there, but I have a tendency to want to attenuate the large signal at the back end of the pre rather than at the smaller signal at the input.
I do not know. The Autoformer is complicated subject for me and would not be the person to talk about it. Year back a personal friend of my made us a very good Autoformer that he was very proud of. In a few years them he sold it to a larger company and it became probably the best performing and the most celebrated/reputed in high-end Autoformer over the last 30 years. As the craze about it grew in 90s I asked the friend of mine to let me to try it. It worked very badly in my estimation but how can you tell it to a friend who was so proud about it? Mind you that I used multi-amping at that time and there is a chance that something in Autoformer use I did not consider as the parameters of the Autoformer because the part of you line-level filter and it became complicated. Also, at that time I use Lamm preamps that were very different animals and they did some very interesting for me phase processing, which no Autoformer could do…. 


"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche
09-03-2019 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 179
Post ID: 25584
Reply to: 25556
Bass Channel OPT
Today, I decided to measure input voltages that induce clipping in my operational DSET.  I basically just put in place the appropriate load resistor, set the signal generator at a level in the middle of the bandpass and increased voltage until the sinewaves showed the first signs of clipping.

Channel A - < 80Hz = 5.3Vpp @50Hz into 1R; 4.9Vpp @ 20Hz into 1R; 4.8Vpp @18Hz into 1R 
Channel B - 60Hz to 500Hz = 10V+ @100Hz into 15R.  Could not clip this channel.




Channel C - Fullrange - 5.3V @ 1kHz into 15R.  

Channel D - 600Hz to 1000Hz = 3.85V @800Hz into 15R [This is a single stage SET and will be run well attenuated from the remainder of the channels]

Channel E - 3.2kHz+ = 8.8V @5kHz into 15R with YO186 DHT tube in play

Channel F - 10kHz+ = not tested, but should be honky dory because it is 10KHz+ only.



All good so far.  The clipping limit for the amplifier seems to be set at about 5.3V for both Channels C and A.  C is lower but that channel is attenuated in use so should not prove to be the limiter.



Channel A clips into 1ohm at 50Hz/5.3Vpp, 20Hz/4.9Vpp, 18Hz/4.8Vpp.  My in-room response gets down to 18Hz and then falls away, so 18Hz is a decent target to shoot for, but I seriously doubt there will be any musical content in that 18-30Hz range at full volume, so nothing should be missed.



Here is the response at clipping for 18Hz which is 4.8Vpp (notice that the 80Hz first order low pass is active):


Bass Channel Spectrum at Clipping 18Hz.jpg




Likewise, this is for a more sedate 1Vpp into 1R:


Bass Channel Spectrum at 1Vpp.jpg
09-08-2019 Post does not mapped to Knowledge Tree
Romy the Cat


Boston, MA
Posts 10,160
Joined on 05-28-2004
Post #: 180
Post ID: 25588
Reply to: 25584
Some perspectives...
Anthony, 
 
Yes, the numbers you got are closed to what I expected they shopped be. Frankly the only concern we have is the Channel A. The in-room response down to 18Hz is good and trust me there is a LOT of musical content in that 18-30Hz. A few things I would like you to understand technically to interpret the numbers you get. 
 
First, your measurements again the “appropriate load resistor” has very little practical meaning. The drivers is a dynamic system that has impedance not resistance and the impedance varies with frequency. So, your measurement against 1R might be accurate for 30.5Hz but might not be accurate at 24.7Hz. So, the proper number at which the clipping might take place would be observable at the real load, or the real driver(s) you use. Then there is a definition of clipping. The clipping is situation when a channel has no power to drive the driver but power is current and voltage. You should not be concerned about juts clipping but you need to assure a SYMMETRIC clipping against you real driver across the whole range of the channel. You need to connect a scope to the channel output while driving your driver and drive a single frequency from a generator and increase the input voltage. Observe the shape of the sinusoid and as the input voltage rise you will see that the top or bottom of the sinusoid will bet distorted, or clipped, or flatted down. The top of the sinusoid is voltage insufficiency and the bottom of sinusoid is current insufficiency (or vise versa, I do not remember already). So, you objective should be to get the absolutely symmetric clipping point when the current and voltage clip at the same time. Then it means that you get out of you amp the max power again your given load. 
 
Second important aspect is that if you deal with an amplifiers then you need to understand when clipping comes from. If might come from many locations. The PS in driver stage, the coupling (in case transformers use), or the output stage PS, or the OPT and so on… The Milq was designed in a way that the any power restrictions are coming from output transformers, the way how SET should be designed and the DSET topology would take care of that limitation. So, the primary focus of your in the given topology should be the only the Channel A output stage. If looks like you beef up enough inductance in you OPT of the Channel A, which is good, not you need to make sure that you can pump power to the output tube. Here is there is another limitation. The 6C33C is indirectly heated tube and they type of tube as the enter to the operation what grid voltage approaches to bias voltage (class A2 or the mode of grid currents)  then this type of tubes do not do so well and they distort rather hard. The direct heated tubes if you feed them with enough current they can push through a little bit but the indirect heated cannot.  You it should be very important to you that in your case your Channel A as the input voltage goes up the Channel stags in Class A1 and the input voltage in the grid of 6C33C does not rise to the rise voltage. In fact, knowing how the Milq is designed this would be the ONLY measurement that I would care as it ejectively demonstrate the efficiency of you LF speakers projected to the acoustic size of you listening room.  I had  at my site a post where I described the measurements I took and posted my measurements in my room.  Here it is:

http://www.goodsoundclub.com/Forums/ShowPost.aspx?PageIndex=1&postID=6057#6057

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche
09-09-2019 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 181
Post ID: 25589
Reply to: 25588
More to do
 Romy the Cat wrote:
Anthony, 
 
First, your measurements again the “appropriate load resistor” has very little practical meaning. The drivers is a dynamic system that has impedance not resistance and the impedance varies with frequency. So, your measurement against 1R might be accurate for 30.5Hz but might not be accurate at 24.7Hz. So, the proper number at which the clipping might take place would be observable at the real load, or the real driver(s) you use.


I agree completely, of course.  I made these measurements with the purpose of checking the input voltage levels that I should aim for with the pre-amp project.  A secondary motive was to make sure the Bass OPT was doing something like the job it is supposed.  To be honest, driving sinewaves at circa 5V into the DSET is LOUD, so I have ordered and now received some appropriate load resistors for all channels so that I can be more thorough with re-testing without making my ears ring.

If I look at the impedance/frequency/phase plot of the Bass Cannons (below), 18Hz is pretty close to the most difficult load that amplifier channel will see.  Most of the passband is in resonance so *should* be easier to drive, however I do plan to re-do the measurements using the Cannons as the load.

DATS First Cannon (Left).jpg



 Romy the Cat wrote:

  
Then there is a definition of clipping. The clipping is situation when a channel has no power to drive the driver but power is current and voltage. You should not be concerned about juts clipping but you need to assure a SYMMETRIC clipping against you real driver across the whole range of the channel. You need to connect a scope to the channel output while driving your driver and drive a single frequency from a generator and increase the input voltage. Observe the shape of the sinusoid and as the input voltage rise you will see that the top or bottom of the sinusoid will bet distorted, or clipped, or flatted down. The top of the sinusoid is voltage insufficiency and the bottom of sinusoid is current insufficiency (or vise versa, I do not remember already). So, you objective should be to get the absolutely symmetric clipping point when the current and voltage clip at the same time. Then it means that you get out of you amp the max power again your given load. 
  


My procedure was pretty much as you described: put a sinewave into the DSET, watched that sine on the oscilloscope (software), change the voltage level to find when the top or bottom or both of the sine starts to flatten.  Not all channels showed symmetrical clipping and I cannot accurately remember how the Bass Channel clipped, but I think it was the bottom of the sine first having problems and will certainly take more notice when I re-do the measurements now that I have more load resistors to quieten the room.

 Romy the Cat wrote:

  
Second important aspect is that if you deal with an amplifiers then you need to understand when clipping comes from. If might come from many locations. The PS in driver stage, the coupling (in case transformers use), or the output stage PS, or the OPT and so on… The Milq was designed in a way that the any power restrictions are coming from output transformers, the way how SET should be designed and the DSET topology would take care of that limitation. So, the primary focus of your in the given topology should be the only the Channel A output stage. If looks like you beef up enough inductance in you OPT of the Channel A, which is good, not you need to make sure that you can pump power to the output tube. Here is there is another limitation. The 6C33C is indirectly heated tube and they type of tube as the enter to the operation what grid voltage approaches to bias voltage (class A2 or the mode of grid currents)  then this type of tubes do not do so well and they distort rather hard. The direct heated tubes if you feed them with enough current they can push through a little bit but the indirect heated cannot.  You it should be very important to you that in your case your Channel A as the input voltage goes up the Channel stags in Class A1 and the input voltage in the grid of 6C33C does not rise to the rise voltage. In fact, knowing how the Milq is designed this would be the ONLY measurement that I would care as it ejectively demonstrate the efficiency of you LF speakers projected to the acoustic size of you listening room.  I had  at my site a post where I described the measurements I took and posted my measurements in my room.  Here it is:

http://www.goodsoundclub.com/Forums/ShowPost.aspx?PageIndex=1&postID=6057#6057


Thanks Romy.  This is the stuff of which I am uncertain.  I'll work through it...I think it is important for me to understand this aspect of the design, and also to evaluate the Bass OPT.
07-26-2020 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 182
Post ID: 25895
Reply to: 25589
I had it wrong all along...
Work has been so crazy busy all last year and through these strange times right now that I have not been having any time to think about the audio project let alone indulge in advancing it.  One DSET/Macondo has been playing mono and I have really been enjoying it.  Then a few weeks ago we took a family holiday to the beach (no C19 community transmission in my state of Australia...restrictions are reasonably loose) and I resolved to try to take a weekend for myself here or there.

The first weekend was spent cleaning and reorganising the room.  It is amazing how much dust gets through things and how long those Bass Cannons take to wipe down properly.  That was a load off my mind.  It felt so good to do something towards the audio.

Then this weekend, even though I should be working, I decided that the time was right to have another go at biasing the second DSET properly.  That week by the ocean had given me time to think about the problem I was having with the amp, and although I had already spent probably more than a week (or two) over the past 18 months trying to get things to work properly, my electronics beginner status seemed unchallenged.  By having the problem leave my head for six months, then revisiting it, where the problem lay became clearer and over the next couple of weeks I snuck moments to think about the method I would use to verify and ammend it.

It worked straight away.  Plus it has led to me finding a problem with the DSET to which I have been listening since March last year.  The first stage of Channel B has been running hot...270V at the anode where it should have been 200V.  Not sure how I could have missed that...but I did.  The fix for that issue is more challenging.

Romy, perhaps you are able to remember some of the lessons from when you were developing these amps all those years ago.  Basically, the problem is that the first stage of Channel B is biased at -7.15V where the remaining channels are in the vicininty of -4.2V.  I can change either the 12.1kR resistor to a smaller value (about 2.9kR) to bring B back into the fold, or I can double the size of the 402k resistor (between the aforementioned 12.1kR and 10uF cap).  Both will work according to my calculations.  I think you are going to suggest to change the 402kR to circa 800kR rather than 12kR to 2.9kR, but the former is much easier for me to do than the latter.

Happy days though!  I am going to let the amp soak in downstairs on the test bench this week and assuming no problems will grab a couple of blokes for the lift upstairs into position in a week or so.
 
07-26-2020 Post does not mapped to Knowledge Tree
Romy the Cat


Boston, MA
Posts 10,160
Joined on 05-28-2004
Post #: 183
Post ID: 25896
Reply to: 25895
Play It Again, Sam
Anthony, I would recommend you go other direction. I very much assure that with over 7V on fist stage bias the channel is dead, it should be around 4V. Let see how to fix it.  I would certainly do not change neither 12.1kR resistor nor 402k resistor. Both of them form loading impedance for the filter and it you change it then you need to move the value of the cap to ground (it forms subtracting low-pass filter at 500Hz). It is not a big deal to do and it might be that you want to move the crossover pose to one or another direction. It is your room and your sound, you can do whatever you want. Still, remembering how much time and efforts I spent to find the configuration that I published I would recommend you to try it once as is and then move to whatever configuration works in your room and for your ears. So, instead to changing the values let think why you have wrong voltages if the PS voltages are as they should be. I am sure that you tried to change the tube but I think it is not the case. I think you out in there a wrong resistor accidently. It is easy to confuse them, and particularly is you use my Dales who mark the values by a position of a dot. If instead of 402k resistor you put in there 40.2k resistor then you will end up with what you have. If you do not want to take the resistors out and to measure them you can measure the resistance by measuring the voltage drop before and after resistor. I am sure that wrong resistor in bias cha is the problem. Even if you do have access to the resistor and can read the proper value on it then take it out and measure, it might be faulty.

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche
07-26-2020 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 184
Post ID: 25897
Reply to: 25896
R4 is the Culprit
Thanks for your help Romy.  Have a look at this (barely legible) sketch so we have a common reference:

Channel B Bias V2.jpg



This is Channel B First Stage bias circuit.  All is identical to your published schematic apart from R9 which I had to increase to get the other stages to bias up around -4V.  In red are my voltage measurements at those points with the current calculated according to the nominal resistor value:  they seem to correlate.  Note that R4 has a voltage drop across it and contributes -2.9V to the bias for the 6E5P.  This is different behaviour to all of the other channels where the crossover filters do not impact the bias for the first stage tube.  Channel B filter is unique in that it has a shunt resistor R4.

If you recall Channel A has foregone the 12kR resistor (R5) and uses a larger 1M R7.  If I did this, I could remove R4 entirely, swap out C1 for 0.01uF and retain the same low pass filter frequency.  It is essentially the same as Channel A but wil a different LPF.  See below:

Channel B Bias Solution.jpg


R9 can then be adjusted to get the ideal bias voltage, and all three channels on this bias circuit (A/B/C) will adjust by the same amounts.
07-27-2020 Post does not mapped to Knowledge Tree
Romy the Cat


Boston, MA
Posts 10,160
Joined on 05-28-2004
Post #: 185
Post ID: 25898
Reply to: 25897
Some explanations
I see, let me to explain how the this work all together the you will be able to make your own judgement. Let look at the functional purpose of the emails in there.  
 
The R1 resistor is a loading resistor. This is understandable. It makes sense to keep it high as we have 5 parallel channels to run that drops the impedance to 1/5.
 
The R2 is the resistor over which the positive bias is suppled, balancing the input to 0V. It is a good idea to keep is as high as possible as it acts as a decouple between the tube and the adjustable resistor.  This DC voltage goes directly were the signal flows what so it should “conditioned” as much as possible. The higher R2 resistor the less the AC signal “sees” the elements behind the resistor. BTW, you have in there missing another resistor and a very small capacitor to the ground over which noises get shunted. This cap very impactful to sound, so you could experiment with it.
 
The R3 and R4 is are a regular voltage divider set to align the output of your mid-bass channel with the rest of the channel. It your room it might need a different adjustment depending on how you room response to the given position of your mid-bass horns.
 
C1 is a low pass filter again the impedance of everything you have on the right.
 
R7 is the same as R2 but for negative supply. There are a lot of missing there as well but you might did it internally to make the drawing dimplier.  Do experiment with that C2, that cap, in particularly on negative supply for some reasons VERY MUCH affect sound. I ended up use an electrolytic cap, which is kind of ridicules but sounded the best to me at that time.
 
R6 is a grid resistor. It is very fast tube with a tendency to rind and to have microphonics, so the grid resistor helps little bit with that. You can put sone ferrite into play that would do the same. When you just turn the amp and it is stone cold you can clearly hear a very unpleasant sound with a lot of ringing. That is from 6E5P grid. BTW, the length of that ugly “glossy” sound high be a good is a good indication for you when to change the 6E5P.

"I wish I could score everything for horns." - Richard Wagner. "Our writing equipment takes part in the forming of our thoughts." - Friedrich Nietzsche
07-28-2020 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 186
Post ID: 25899
Reply to: 25898
Fixed
Thanks Romy,

It is now fixed.  I have left Channel B the same apart from adding a 390k positive bias resistor on the positive signal line (between C1 and R5) connected to the 14k resistor that also connects to Channel C and F.  A 402k would have been used had it been on hand.

The solution came about by thinking about simple things that Romy could have transcribed when preparing the schematic, and it worked right away.  The beauty of this solution is that changing the attenuation for Channel B (R4) will not alter the bias voltage by much and also there is much less drop across that resistor from the negative signal line.  

Now, all the channels can be set to about -4.3V bias so it is working.

Regards,

Anthony
07-30-2020 Post does not mapped to Knowledge Tree
anthony
Posts 338
Joined on 08-18-2014
Post #: 187
Post ID: 25900
Reply to: 25899
Two up
Impatience got the better of me, and I called two mates who prompty showed up yesterday afternoon and we hefted the second DSET into place on the horn stack.  A couple of hours of connecting and checking things and it was all ready to fire up for my first sterophonic sound in 18 months.

2up1.jpg


What a delight!  Even with a dying 45 in the right channel and a sweet YO186 in the left!  Especially with the lovely Helix Two/Schroder CB/MC Anna in action.

Helix2.jpg



The finish line is getting close now.  Have t find some time to finish the lathe work for the front half of the upperbass horns and then it is general tuning and room treatment.

Anthony 
Page 8 of 8 (187 items) Select Pages:  « First ... « 4 5 6 7 8
Home Page  |  Last 24Hours  | Search  |  SiteMap  | Questions or Problems | Copyright Note
The content of all messages within the Forums Copyright © by authors of the posts