Joachim, using Burmester as your reference point may wll be in line with a point you may be making, but because of its prohibitively high price, it can hardly be a general reference. Far too few people have the money to buy it.
There were a few cheap CD players which did well, in fact much better than they were expected to do, and their manufaturers withdrew them because they sounded so good nobody really wanted their more expensive models. Philips 720 comes to mind. At around DM 120, it made music which was slated for models costing over DM 400 in its day, in the late 90ies. My point is, as a customer, one also faces orporate strategies and tiers. I still keep one for memory's sake.
My own Dual CS 604 turntable is another good example. After it appeared in 1977 (I think?), German audiophiles discovered that it sounded almost as good as the much more expensive CS 701, so a list of customers was made, I had to wait two months for my sample to be made. Of course, the sales of 701 dropped like a rock, so Dual killed the model about 5 months later with its replacement, model CS 606, which restored their "natural order".
Dejan you have big advantage in your amp. It is super tunable. Loop gain, VAS degeneration, VAS Cdom. The reason I mention VAS degeneration is relaxing the feedback loop will loosen the damping factor. Making the VAS degeneration less will restore that mostly. As the loop gain increase the need for Cdom reduces. You might find nearly every thing is as before except a greater sense of space. Obviously HF distortion must not be allowed to rise. Cdom should counter that. The bass distortion will change. Not so much as to be as the HF distortion. You should find it drives nasty speakers even better if so. There is even two pole Cdom if brave. It seems to work.
One of the oldest questions in the book - how much damping do we actually need?
Regarding my case, let me remind you that the worst case will contain three pairs of output devices, my sample four, which in itself improves output impedance.
Next, I plan to use rather large toroidal transformers, another good sign for a low damping factor, in a dual mono configuration. Ditto for electrolytic capacitors, I always prefer to use two 10,000 uF caps in parallel than one 22,000 uF cap.
Lastly, I will use two paralleled output relays, cutting their added series impedance, small as it may be, in half.
All told, I don't think I have a problem in the damping factor department.
On the other hand, after Demian's help, my latest effort (still in purely simulator form, of course) hits 115 kHz at full blast, open loop, and into 4 Ohms. Since it has been constructed to be relatively easily adjusted to various needs, I can always: a) increase the value of the Miller caps, b) increase the degeneration resistor value in the VAS, or do both inn a balanced form. The point is, I have built-in reserves to increase the NFB factor to almost anything I like because I can either degenerate or clam down the VAS almost as I please.
Just what to do will have to be determined after I make a test model and see how it behaves in real life. It's true my simulator is rather conservative, so live models tend to do better than it says they will, but still, a live sample is a live sample, and nothing beats that. If the 'scope confirms what the model says, I'm home and dry. We'll talk more about this in good time, Nige.
Regarding my case, let me remind you that the worst case will contain three pairs of output devices, my sample four, which in itself improves output impedance.
Next, I plan to use rather large toroidal transformers, another good sign for a low damping factor, in a dual mono configuration. Ditto for electrolytic capacitors, I always prefer to use two 10,000 uF caps in parallel than one 22,000 uF cap.
Lastly, I will use two paralleled output relays, cutting their added series impedance, small as it may be, in half.
All told, I don't think I have a problem in the damping factor department.
On the other hand, after Demian's help, my latest effort (still in purely simulator form, of course) hits 115 kHz at full blast, open loop, and into 4 Ohms. Since it has been constructed to be relatively easily adjusted to various needs, I can always: a) increase the value of the Miller caps, b) increase the degeneration resistor value in the VAS, or do both inn a balanced form. The point is, I have built-in reserves to increase the NFB factor to almost anything I like because I can either degenerate or clam down the VAS almost as I please.
Just what to do will have to be determined after I make a test model and see how it behaves in real life. It's true my simulator is rather conservative, so live models tend to do better than it says they will, but still, a live sample is a live sample, and nothing beats that. If the 'scope confirms what the model says, I'm home and dry. We'll talk more about this in good time, Nige.
Last edited:
Impressive THD and noise..... If you need a gain of one circuit.
THx-RNMarsh
Richard, just trying to understand, why do you say that? Why don't you think these devices could be used at higher gains?
Damping guesstimate is > 16 ( very old statement of Hi Fi Choice). It can be as low as 3 and do something (Wireless World experiment 1955 for gathered edge surround).
I favour semi active speakers. That is you design a speaker with minimal crossover and buy a dedicated amplifier when money permits. That way the freedom of design is maximum. I designed a speaker like this so that NAD and Rotel amps could provide good performance. The tone controls would be used. Much harder than you would think. All Quad and Sugden amps also. Ultra minimalist amps would be fine and would be no worse than Linn Kans of old. The sound of a speaker like this although not invisible is very like real music in the slightly less good seats. With the dedicated amp the better seats become available. They are miniscule which will be liked. Modeled on the weakness and strengths of Quad ESL 57. That is they have a sweet spot stereo image and the bass is about the same in quantity. A Rotel 931 and these speakers really says what the singer is singing about and is so on 78's also. In many ways the Martin Logan and Krell I heard was far behind it. Superficially the later was more refined ( by a long way in truth). My little set up is real artists in a slightly bad room. I can live with that. With the dedicated amp we change the room.
I favour semi active speakers. That is you design a speaker with minimal crossover and buy a dedicated amplifier when money permits. That way the freedom of design is maximum. I designed a speaker like this so that NAD and Rotel amps could provide good performance. The tone controls would be used. Much harder than you would think. All Quad and Sugden amps also. Ultra minimalist amps would be fine and would be no worse than Linn Kans of old. The sound of a speaker like this although not invisible is very like real music in the slightly less good seats. With the dedicated amp the better seats become available. They are miniscule which will be liked. Modeled on the weakness and strengths of Quad ESL 57. That is they have a sweet spot stereo image and the bass is about the same in quantity. A Rotel 931 and these speakers really says what the singer is singing about and is so on 78's also. In many ways the Martin Logan and Krell I heard was far behind it. Superficially the later was more refined ( by a long way in truth). My little set up is real artists in a slightly bad room. I can live with that. With the dedicated amp we change the room.
It works very well and has a small advantage over negative feedback that produces similar cancellations as the minute phase shift isn't there. Can be used with negative feedback to be even better if chasing measurments. My pentode in 82% UL out amp works very well with it. Tested with a bucket of worn out valves to be sure it would work over many years. 1 % THD at 5 watts my target with exponential harmonics using zero loop feedback ( zero any type of negative feedback as far as I can see as both cathodes are bypassed and no Schade/Kitic shunt). The UL is because it still looks like triode and I get a small bit of gain ( 30 % ). You could say the triode has some internally as it is an EL34. If I add 20 dB loop feedback it looks rather good. Same amp without pre-distortion would be 5% THD ( all triode ). That is 14 db more. I noticed that the amplifier was happier to accept negative feedback also. No idea why. To top it all the pentode has 14 db extra gain. That starts to look decisive. People said I was fooling myself. My ears said they were wrong as it sounded like it measured only much better. Very open and no ringing. Ringing in valve amps sounds nasal.
Sorry, Richard, I missed this comment earlier - the idea was ventured by me as a "proof of concept" more than anything else, on the basis that by iteratively pre-distorting and analysing the results that highly accurate reproduction could be achieved - and a lot of the distortion to be countered would be contributed by speaker drivers. Let's say, we want a train of high quality 60Hz square acoustic waves to be produced; this would most likely entail quite a number of rounds of tuning the pre-distortion, to account for the drivers altering their behaviour in not particularly predictable ways as the input is varied, but in each run getting closer to the desired acoustic waveform.
I prefer MP3. It promises far less and sometimes better. Someone said to me MP3 not bad as it is 3% of the original. Then he said ironically CD is 10%. Possibly that is the explanation. It has the ring of truth to it. If we say the last square-wave if CD is about 1kHz that amounts the the same thing.
Digital gets a lot of bad press because it demands so much of the replay mechanism, which often is too sloppy in some area - hence, somewhat undesirable sound. Get it right, and then it all falls into place, you realise how most audio sound is out of whack, well short of what it should be ...
MP3 is too noisy for playback on a hi-resolution system, same for its limited dynamics redbook is vastly superior , very much so when played back on CD player VS PC , we have done the many, many comparisons, cd direct playback beat PC COPY VIA SAME DAC , pc only superior when doing hi-reso playback .
I think I know what you mean.
Sometimes, old is much better than new. I would probably be running an oldie, but the problem with them is that if the drive dies on you, you can no longer buy a replacement new drive for almost any model by anybody. These days, only combos (CD+DVD or CD+DVD+Blue Ray) seem to be made, and by looking at them, my feeling is they are getting worse and worse in mechanical terms.
The days of the indestructible CDM-1 are gone forever.
My thinking exactly.
In these days of cheap memory sticks, a 16 GB stick is very affordable, while 4 GB sticks are dirt cheap. 4 GB is roughly equivalent to 5 Red Book CDs. That's cheap enough for me to use them rather than actual CDs even in my car.
In other words, who needs compression, except extremo cheapskates?
Today I was looking up alternatives for comparators that can both source and sink current. A NE 555 is not so good because the window can not be set to the miniscule difference I was looking ( better at low voltage ). LM339 isn't much use as I would have to do some work on it to get it to function. My old friend LM324N came to mind. 22R and 10 K to give it hysteresis as a start ( output to + input with - input to control voltage). I will use the 3 spare sections as a source/sink buffer. The voltage reference TL431 at 1.5 mA. LM324N looks much like a comparator and has some performance advantages over other op amps in this application.
It might interest people to know this is inspired by a piece of Indian power control gear given to me to investigate. Remarkably simple and effective. They are so careful with money they use the LM324N in their circuit as a current sink. This gives the reference voltage via a resistor !!!!. I found it worked rather well and found it was OK with a group of different make 324's. The other voltages created by a voltage doubler that has a centre zero. The window comparator a string of resistors that look to have been arrived at empirically. My version which is 90% different is far simpler and can be calculated. I have a feeling it will get made.
In my search for other ideas I came upon this. Like me they guy forced an LM324 to behave ( see graphs ). I used a RA53 thermistor instead. I also used a CCS . When I did it equaled a TL072 and a NE 5532 up to 22 kHz. Whilst some would question the value of this the LM358 was a 10 second substitute. It also proves without doubt that class A CCS works. These tricks when used with high grade op amps are subtle. I always think as long as not inferior give it a try as long term it might give less listening fatigue. The slew rate of 324 is 0.4V/uS. When in class A it might be of less importance. I would never advocate 324 in Audio. That isn't the point, it is knowing how things work and seeing that often they don't.
Budget Wien bridge oscillator with low THD | MyElectrons
It might interest people to know this is inspired by a piece of Indian power control gear given to me to investigate. Remarkably simple and effective. They are so careful with money they use the LM324N in their circuit as a current sink. This gives the reference voltage via a resistor !!!!. I found it worked rather well and found it was OK with a group of different make 324's. The other voltages created by a voltage doubler that has a centre zero. The window comparator a string of resistors that look to have been arrived at empirically. My version which is 90% different is far simpler and can be calculated. I have a feeling it will get made.
In my search for other ideas I came upon this. Like me they guy forced an LM324 to behave ( see graphs ). I used a RA53 thermistor instead. I also used a CCS . When I did it equaled a TL072 and a NE 5532 up to 22 kHz. Whilst some would question the value of this the LM358 was a 10 second substitute. It also proves without doubt that class A CCS works. These tricks when used with high grade op amps are subtle. I always think as long as not inferior give it a try as long term it might give less listening fatigue. The slew rate of 324 is 0.4V/uS. When in class A it might be of less importance. I would never advocate 324 in Audio. That isn't the point, it is knowing how things work and seeing that often they don't.
Budget Wien bridge oscillator with low THD | MyElectrons
Dont be so hard on Franky and English ...
Come again?
Why am I being hard on anybody?
Frank loves PCs and wants good sound from them. I admit to being sceptic about him getting there, but it is a worthy cause.
Why the English? What can possibly be wrong with a nation which has kideny pies, best biscuits this side of the galaxy, outstanding jams and marmelades, phenomenal cider, and so forth? Or were you referring to Nigel?
If so, I must admit some of his points are well worth considering. For example, his story about Bob Stuart saying that each transistor in the signal path is a little liar definitely has its merits. It got me thinking, so in my (rare) spare time, I am trying to simplify my own creation which Nige seems to like, my amp, in an attempt to get there with as little loss of performance as possible, but with as similar effect as possible.
Live and let live, you might be surprised in the end.
In someways Pink Floyd Atom Heart Mother Mother works better. Here the microphone is too close sometimes. I guess because I listen a lot these sounds are everyday.
Joachim. Does placing a choke in series with lets say a mid-range unit reduce the amplifiers ability to damp a HF resonance ? I am in two minds about that as it can grab the unit at LF. For example 200 uH increased to 800 uH or when active near to 0 uH. The 200 uH to take the edge off and the 800 uH someone else recommended that, to my ears it was worse. I see this a lot in speaker design. Take a drive unit that can do 8 to 10 kHz and chop it at 1.5 kHz to make a nice 1960's speaker. When told it was to avoid beaming I said Quad ESL 57 beam ( far worse ). Surely to cross as high as possible has to be best ?
Nige, to quote Bronsky Beat: It ain't necessarily so ...
What you are proposing is very much alike to what JBL did with their Century 100 and 4312 monitors, they cut their bass at around 1.5 kHz and their mid at around 5 or 6 kHz.
However, AR cut their bass at 500 Hz and their mid at 5 kHz, and it still made up for a hell of a good speaker in its day.
I think what will be determining this are the properties of the bass and mid drivers. If you midrange driver can go as low as say 300 Hz, you can cut it at 600 to 800 Hz no problemo, assuming your bass driver will go as high as say 1 kHz. Ditto for high frequencies.
My own midrange driver (Son Audax, as are all other drivers) is rated for 300 Hz on the low side and up to 6 kHz on the top side. My bass driver is rated for 1.2 kH, and my tweeter for as low as 2 kHz. My crossover points are thus placed at 800 and 3,500 Hz, to make sure that the drivers are still well within their linear operating range. XOs are 12/18 dB/oct.
Given that the frequency response is 40-18,000 Hz +/-1.5 dB. I'd say we got it right. This is as good as a professional tape machine at 15 ips. But this is so only with THAT set of drivers, in THAT enclosure.
What you are proposing is very much alike to what JBL did with their Century 100 and 4312 monitors, they cut their bass at around 1.5 kHz and their mid at around 5 or 6 kHz.
However, AR cut their bass at 500 Hz and their mid at 5 kHz, and it still made up for a hell of a good speaker in its day.
I think what will be determining this are the properties of the bass and mid drivers. If you midrange driver can go as low as say 300 Hz, you can cut it at 600 to 800 Hz no problemo, assuming your bass driver will go as high as say 1 kHz. Ditto for high frequencies.
My own midrange driver (Son Audax, as are all other drivers) is rated for 300 Hz on the low side and up to 6 kHz on the top side. My bass driver is rated for 1.2 kH, and my tweeter for as low as 2 kHz. My crossover points are thus placed at 800 and 3,500 Hz, to make sure that the drivers are still well within their linear operating range. XOs are 12/18 dB/oct.
Given that the frequency response is 40-18,000 Hz +/-1.5 dB. I'd say we got it right. This is as good as a professional tape machine at 15 ips. But this is so only with THAT set of drivers, in THAT enclosure.
D,
It was a rhetorical response ...
@Nige,
How are you determining high as possible, why would anyone want beaming, not to mention the added thd if the driver cant handle appropriate spl with such a wide bandwidth. The lack of coherency from beaming is also a big problem , quads never suffered from such , the 63 took care of most of that too ...
It was a rhetorical response ...
@Nige,
How are you determining high as possible, why would anyone want beaming, not to mention the added thd if the driver cant handle appropriate spl with such a wide bandwidth. The lack of coherency from beaming is also a big problem , quads never suffered from such , the 63 took care of most of that too ...
Quad 57 were notorious for having a very narrow window . Some say never a true stereo speaker. Logic of that a little hard to follow as mono requires a better polar dispersion. My SMGa not better than 57's on that.
The OB 12Lta very good considering no one would expect that.
As I said before if no worse than 57's I will be happy. In fact what I built was far better. 63's did address the problem. Some said 63's sound off axis wherever you sit.
The OB 12Lta very good considering no one would expect that.
As I said before if no worse than 57's I will be happy. In fact what I built was far better. 63's did address the problem. Some said 63's sound off axis wherever you sit.
- Status
- Not open for further replies.