@rollintubes
I am sold on tubes for analogue audio but am confused by all of the information on power. I see from many posts that tube power need not be very high or as high as the speaker manufacturer claims as a requirement, i.e. a 200 WPC SS amp is needed to drive a speaker with 85db sensitivity (the manufacturer requires a minimum of 75 WPC, but likes at least 100 WPC), yet I have used a tube power amp with 40 WPC on the speakers and it sounds terrific. I have read that it is in the output transformers and SS amps are generally direct coupled.
Will you please explain this phenomenon?
It is my understanding that 200 watts per channel is a maximum to not blow up the speaker. That is a lot of power for a voice coil to absorb. Almost everyone has too big an amplifier in my opinion. I know this as I have measured power at listening levels in customers homes to be only a few watts. To say 200 watts is "needed to drive a 85 db speaker" means that the speaker will be putting out 105 db SPL. I listen at 85 db, anything higher is for short periods. Do you listen at 105 db? Your 40 watt tube amp sounds terrific because thats all the power you need. Small amps generally sound better than big amps. In designing big amps, certain sins are committed. Small amps can be more responsive and delicate. I am currently designing several 30 watt/channel amps for those who agree with what I have said above. |
http://www.audiosystemsgroup.com/TransLines-LowFreq.pdfThe last paragraph of this 3 page paper dealing with transmission lines admits that we are not in transmission line territory at audio frequencies..
Electrical Wavelength at Audio Frequencies can be computed from Fig 2.
At 1/20 wavelength and below, transmission line effects are negligible, and
the circuit can be analyzed using simple
lumped circuit parameters. At 1/10
wavelength, transmission line effects are
still quite small, and can generally be
neglected. Fig 3 shows the characteristics of the RG59 cable of Fig 2. Typical
balanced audio cables have similar
properties, typically differing from Fig 1,
2, and 3 by no more than a factor of 2
in frequency. Thus, at audio frequencies, a cable less than 2,000 ft long is no Fig 3 Electrical Wavelength at Audio Frequencies
more complicated than its series resistance and parallel capacitance. As the cable becomes
longer, or as frequency increases, the cable will begin to behave as a transmission line.
michaellent
Why can’t people stay focused? Ramtubes started this discussion about amplifiers and it looks like it fell into the cable whole again!
Start your own thread about cables of any different kind and stay out of the ones that have nothing to do with it! Thanks Michael. I appreciate your request for a return to amplifier discussions. |
the velocity of propagation of the signal (versus the velocity of the actual electrons) is determined by the dielectric or insulation material that the electromagnetic wave is predominantly traveling through.
I would like to see some proof of this. |
@michaellent
I want to learn about amplifiers
Hi. What would you like about amplifiers? They are my favorite subject and life's work. Did you check out the YouTube videos and Burning Amp? Realizing early on that the applications in the RCA and other manuals are just one of many possibilities I like to create new applications that are within the capabilities of the tube yet not in the book. Most USA manufacturers copy the same "typical" application from each other. European makers often have many more applications. I noted this in my BAmp talk. The RM-10 is unique in that it gets over 40 watts (at the plates) with one pair of EL-84s. In the standard application EL-84s achieve 18 watts per pair. However that application is based on a low B+ which is desirable if one wants to make highly cost effective amplifiers. So most manufacturers use that application though many other applications could be created. Mine is at a very high B+ which I have found to be very reliable and gives longer tube life than some of the low B+ apps. Emission Labs, makers of large triodes like the 300B, lists many applications (operating points, load impedances, etc) for each of their tubes. I did some for Jac Music on the new 45B and confirmed them on the bench, not by simulation. See note below applications. http://emissionlabs.com/datasheets/EML45B.htm |
@radiointerference
Hello Roger, thanks again for this opportunity seeing you worked for Harold Beveridge , I have two questions, 1). I have a pair of Beveridge 2SW's with a panel that has gone south. I am looking to make a jig to tension the mylar film. Can you shed some light on how to rebuild the panels, I have some background info on the mylar but would like to know the tension techniques to properly rebuild. We stretched the mylar a bit beyond its elastic limit, so tight as you can get it. A silk screen stretcher works well or look on YouTube for other clever ideas. Weights will not do the job. We have the mylar in stock. It has to be coated both sides and heavy and well attached to the brass strip. There is significant current on the mylar. Nothing like the high resistance film of a Quad. Question 2), I had purchased a set of Counterpoint SA-4's about 2 years ago needing repair/restoration. One was functional one was not but found one tube had vented, I think it was due to mishandling during transportation.
Can you give info on what to check for, and possible ECO's on this amp so I can proceed with a restoration for a set of Quad 63's I am planning to use them on.
I dont have any ECOs. I gave Counterpoint the design and Michael Elliott would be the one who changed things. Dont get his individial bias mod. Try to make the amps as original as possible. We have tubes for them. You need a schematic and some good diagnostic skills to get them going. |
@c1ferrari
Re: Beveridge Hi, Roger,
You'd worked on the Bev DD amps for my Model II's when you were situated in Santa Barbara. Unfortunately, Rick was unable to complete the restoration process and my panels require attention. Can you assist?
Thanks for the attention.
We havent done any panels for a long time. Chat with the other fellow on here who wants to stretch some mylar. We have the right Mylar as mentioned in my reply. |
@jyprez
yes, I would be happy to pay for coursework on the subject. What do you offer?
Currrently basic amplifier design. Are you local to Richmond. CA 94806? |
ON TONE CONTROLS
The typical bass control has a hinge point at 1,000 cycles, much too high. In my system I use the Subwoofer level control (100 Hz and below) to set the bass to the right level without making vocals muddy like most tone controls do. Typical bass controls have lift the male vocal 3 dB to get 6 dB in the low bass. We dont want to hear that.
My woofer level control is right next to my volume control on my crossover and I note I move it several dB depending on the recording, listening level and bass quality.
For treble, usually I want a cut for bright recordings, rarely would I want a boost. A cut control is much easier to make than a boost. Typical cut/boost controls have a hinge at 1,000 also which is too low.
Whoever chose 1 KHZ as the hinge point made a big mistake, and once made was copied over and over. I would choose around 200 Hz for the bass and 4 Khz for the treble and leave 200-4,000 unmodified. |
Concerning tone controls, thank you,,, very good to know. |
Concerning tone controls, thank you,,, very good to know It appears in design the good ideas are only done briefly by the few yet the bad ideas are done by the many and go on forever. I think the choice of a single hinge point was the biggest mistake. Its actually easier to split them in two which I did in the C-4 preamp. |
Roger, could it be the old 1kHz frequency was chosen in relation to the RIAA equalization curves? |
@bdp24
Roger, could it be the old 1kHz frequency was chosen in relation to the RIAA equalization curves?
That would have been a better choice resulting in 500 Hz for the low hinge and 2 Khz for the high. I would prefer a larger gap as stated earlier. The single point of 1 KHZ is trying to hit the middle numbers wise but not sound wise. The RIAA points were chosen by the density of program material. BTW the RIAA curve we are familiar with that is 40 dB of EQ from top to bottom is not the RIAA curve. It is the RIAA plus MM/MC curve. The RIAA playback curve for a semiconductor cartridge is only 12 dB from 500-2,000 Hz. Thus RIAA boosts the highs 12 db and cuts them 12 dB on playback... Does anyone find this interesting or alarming that we call the RIAA curve the big one and not make much mention that it is for velocity transducers only? |
BTW the RIAA curve we are familiar with that is 40 dB of EQ from top to
bottom is not the RIAA curve. It is the RIAA plus MM/MC curve. Try telling the manufacturers of cutter electronics that. Does anyone find this interesting or alarming that we call the RIAA
curve the big one and not make much mention that it is for velocity
transducers only? No. Its well understood that is the case. There has been some controversy over other types of transducers as you point out, won't be properly equalized. IMO/IME the producers of such cartridges should offer their products with an equalizer to set things right. |
Hey Roj, Its Verastarr Mikey..
Id like to know when we roll tubes and hear differences that are so different and unique even when its the same exact tube type, how and what exactly are we hearing ? Lets take 2 6sn7GT for instance, that for the purpose of this question have near identical electrical characteristics.
Thanks man !
|
@atmasphere
Try telling the manufacturers of cutter electronics that.
How about telling us about the cutting process? The range of constant velocity cutting and range of constant amplitude cutting. I was not discussing cutting, but you appear to have something to say. I simply wanted to inform the readers here that the RIAA curve they generally see is not the RIAA curve but the RIAA plus EQ for velocity cartridges. From what I read the Amplitude cartridge people still have not caught on.. |
Hey Roj, Its Verastarr Mikey..
Id like to know when we roll tubes and hear differences that are so different and unique even when its the same exact tube type, how and what exactly are we hearing ? Lets take 2 6sn7GT for instance, that for the purpose of this question have near identical electrical characteristics.
Thanks man ! First how many of the characteristice are near identical. Mu, Gm, Rp and bias point. These vary widely. With 6SN7s microphonics can play a large role. I cannot comment on what you are hearing. If you want to be sure of what you are hearing it takes two identical circuits for the two tubes, an A/B switch, matched levels. We find with many listeners that the difference we hear rolling tubes are small enough to disappear. One cannot just swap a tube, we have found the sonic memory to be too small. I did a strict A/B for 50 people in the SFAS. They looked at me and said. Hey these all sound the same, where are the differences. Given that they want differences, we are repeating the test this saturday and will not A/B. It takes away all their fun when thinigs sound the same Sorry, but good equipment sounds good. Because I test a lot of tubes, use them in off book applications I get to see some real measurable differences but then the tubes are actually different on the test gear too. A curve tracer will tell you a lot. The question I have is why do you want these differences? If you could design your own preamp wouldnt that be more rewarding than swapping parts? |
The LP cutter head is fed a signal powerful enough to allow it to cut a groove on the surface of the lacquer which is an aluminum disk coated with lacquer.
The amps that drive it are just regular power amps although they are typically bandwidth limited to 40KHz or so. It is impossible to clip the amps; they can usually make about 10x more power than the cutter head can withstand.
The signal sent to them has the RIAA pre-emphasis. It is the inverse of the RIAA playback equalization curve. No other EQ is applied (although typically there is compensation for individual cutter head response, so the electronics are trimmed to match the head with which they are used).
So there you have it.
|
@ atmasphere
So what is the cutting curve? Can you show us please? |
@atmasphere
Does anyone find this interesting or alarming that we call the RIAA curve the big one and not make much mention that it is for velocity transducers only?
Ralph please explain your comment below. No. Its well understood that is the case. There has been some controversy over other types of transducers as you point out, won't be properly equalized. IMO/IME the producers of such cartridges should offer their products with an equalizer to set things right.
Would readers care to comment that this is news that the RIAA curve is only 12 db not 40 db bottom to top. Producers of amplitude cartridges do provide EQ, they have to in their box that powers the cartridge. Sao Win tried to EQ mechanically with little success. I built the equipment so he could measure his cartridges (which he did not before). Soon after he gave it all up from frustration of lack of uniformity. A year or two ago the Panasonic cartridge was reviewed with its"box" and there were great errors found in the resulting frequency response. The reviewer was a bit disappointed that these people could not get it right. |
There have been a few other strain gauge cartridges offered with similar issues. So what is the cutting curve? Can you show us please?
Yes. The RIAA pre-emphasis is the algebraic sum of the ordinates of three curves, each expressed in dB: A parallel L/R network with a TC of 3180 milliseconds A series RC network of 318 milliseconds A parallel RC network of 75 milliseconds From top to bottom the range is a bit over 35 dB. |
@atmasphere
Ralph, How about a curve. People here cant understand what you said, can anyone?
Can you ever give a simple answer?
From 500 Hz to 2000 Hz the response on the record is lifted at a rate of 6 dB/octave for two octaves. Thats 12 db total. Is this your understanding? |
I'm sure someone can understand my prior answer. I know you can, therefore to your first question, 'yes'. To the second, also 'yes' but the question you asked didn't have a simple answer- some questions are like that :) To the third, a correct answer requires a correct question- the **response** is not raised; the **pre-emphasis** is, and by about 5 db (so the simple answer is thus 'no'). Remember that the curve is just that and is the amalgam of three timing constants. The range of frequencies to which you refer is the area where the response is relatively flat compared to the rest of the curve. It isn't flat because of the ordinates of the timing constants. So its a gentle curve in that range, and the **pre-emphasis** is 'lifted' by about 5 db over that range, not 12. There are plenty of images on the web. Fig. 2 of the article at the link below is pretty good- you can see the 5 db or so rise in the pre-emphasis in the range of frequencies about which you asked: https://www.stereophile.com/features/cut_and_thrust_riaa_lp_equalization/index.html |
Ralph Thanks for the link. It is clear to me that above and below 1 KHZ there is 6 dB of EQ on either side of the red line, making a total of 12 dB as stated before. To be clear I am not talking about what is fed to the velocity sensitive cutting head but what is actually in the groove and what one must compensate for with a constant amplitude cartridge. It is interesting to note that in the better RCA consoles of the 1950s RCA actually applied this 12 dB shelf to their crystal (amplititude sensitive) cartridges. I dont know anyone else who did. The time constants are indeed 2 octaves apart at a rate of 6 dB/octave. Of course the corners are rounded but the eventual preemphasis is 12 dB starting at 500 Hz. See figure 2...
https://www.stereophile.com/features/cut_and_thrust_riaa_lp_equalization/index.htmlThis is also interesting. https://en.wikipedia.org/wiki/RIAA_equalizationSince the readers here dont seem too interested, I am going to sit this one out. |
Thanks for the link. It is clear to me that above and below 1 KHZ there
is 6 dB of EQ on either side of the red line, making a total of 12 dB as
stated before. I think you might be misinterpreting something- The red line is not the EQ curve. If you look at the actual curve, at 500Hz it only rises about 3 dB to 1Khz and then about 2 1/2 dB to 2KHz. The scale on the left is 5 dB per gradient. I've seen a number of older console and stand-alone components that had networks installed for the ceramic cartridge inputs. Loading of ceramic cartridges is pretty critical if you wan them to work right. Strain gauge cartridges seem to be the high quality version of a constant amplitude device in that they can track at 'normal' tracking pressures and have decent bandwidth. But its always bothered me (nor am I alone in this by any means) that they are often sold without proper compensation. I went to hear one back when CES was still at the Alexis Park and brought an LP that I knew quite well since I recorded it. You could tell it was the same music but the tonality was definitely off. The manufacturer was claiming that it needed no EQ... I thought about offering to do an equalizer for it but with his remonstrations it seemed an uphill battle. I think your comment about the Win hit the nail on the head- there's a bit of a consistency problem. |
@atmasphere
µs, not ms
75 µs, 318 µs and 3180 µs
|
µs, not ms
75 µs, 318 µs and 3180 µs <facepalm>Correct!</facepalm> |
Sir---- I cannot get any responses on the Digital forum to my query about the variables affecting the implementation of dac chips in dac units. So, I hop that you can address this question for me. [My private email address is [email protected] if you prefer to address it is private.] Here is my formation of the question: What makes for good implementation in a dac? By now it is conventional wisdom on this digital site and others that the dac chip itself counts for comparatively little of the overall quality of the unit. The slogan has become something like: "The chips 10%, the implementation 90%." However, I cannot recall any very detailed discussion of precisely what very good to excellent to superb implementation actually involves.
Yes, there are general references to the quality of the power supply and especially to the quality of the analogue output stage; but nothing [that I can recollect, anyway] that goes deeply into the details....
So, I raise that question here. I do hope that the technologically adept members of this forum can address it, with some attention to the various dimensions of implementation, e.g. op amps vs tubes in the analogue output stage, some concrete examples of brands and models that do it well and do it badly. |
I can say from Roger's perspective that while he listens to digital sources, I am not sure he would comment on their circuit designs given his lack of interest in manufacturing such a unit. Roger is all about research first and foremost, as a great designer should be. From my conversations with him I don't think he has done much of the research necessary to formulate a response to your questions. Of course I could be wrong and I'm sure others with an opinion might chime in.
I for one, having tried both, always felt that tubes really don't add much value in digital sources. Yes I'm sure the "it's all in the implementation" folks will jump all over that one, perhaps the "jitter" folks as well, but that is just my experience. I've been using the same DAC for 12 years now and have had others in for audition, but the incumbent stays. Is it the PCM 1704 chips? The battery powered output stage? The direct coupled design? The Word Clock output? Not sure, but all I know is nothing else makes me want to replace it.
|
I cannot recall any very detailed discussion of precisely what very good to excellent to superb implementation [of a DAC] actually involves.
Yes, there are general references to the quality of the power supply and especially to the quality of the analogue output stage; but nothing (that I can recollect, anyway) that goes deeply into the details.... IMO a major reason that the discussions you referred to have not delved very deeply into the details is simply that the details that are involved in the design of a high quality DAC, and the opportunities for the designer to overlook subtle issues that can adversely affect performance, are so vast in number that it would be impractical to address them in anything resembling a comprehensive manner. And it would be misleading to single out just a few of those details for discussion, while overlooking countless others. That is of course true to some extent in any sophisticated electronic design, but it is especially true in the design of a component that encompasses high speed digital circuitry, D/A converter circuitry, and analog circuitry all in close proximity. One major variable that usually seems to be overlooked in such discussions is the criticality of the design of the printed circuit board itself, including where the chips are placed, how signals are routed within the board, and how power and analog and digital grounds are distributed and "decoupled" (loose translation: "kept pure"). Take a quick look at the Table of Contents of the book "High Speed Digital Design: A Handbook of Black Magic," written by a noted authority and consultant on that subject, and at some of the pdf’s linked to in the "Downloads" section near the bottom of the latter page. You’ll get a small idea of the complexities that can be involved in the design of purely digital high speed circuits. Add D/A converter circuits and analog circuits into the mix and the opportunities for a design to become less than optimal grow dramatically. As the saying goes, the devil is in the details. And IMO what usually accounts for much and perhaps most of the difference between very good and excellent and superb implementation is simply the knowledge, expertise, and experience of the designer. Regards, -- Al |
One major variable that usually seems to be overlooked in such
discussions is the criticality of the design of the printed circuit
board itself, including where the chips are placed, how signals are
routed within the board, and how power and analog and digital grounds
are distributed and "decoupled" (loose translation: "kept pure").
My take on it is layout. A single misplaced trace can shoot down your design because of the noise it can make. And Al is correct- this is merely the tip of the iceberg. |
I 100% agree that there are many more excellent sounding lower power amps than higher power amps. Most appear to be between 20 and 40 watts. I have a voltage regulated (non-ultralinear) 35 watt Dynaco 70 that is superb sounding. I have also have the Macintosh MC 30s, Yamaha CR 620 and have heard the Macintosh 225 and 240 and several Marantz 1970s receivers. Some favor the tubey sounding Fisher 400 and 500 receivers (I own them but don't find them as resolving as the above mentioned Dynaco, Macintosh and Yamaha units). |
Roger considers the Marantz 8b superior to the Model 9. The Model 9 is sure cool looking, though. The old-style Atma-Sphere faceplates remind me of the Model 9's (except the Atma's had ultra-cool champagne anodizing, my all-time favorite). Coincidence, Ralph? ;-) |
No- I liked the styling of the Marantz. I was also influenced by ARC. Both of these can be seen on the front panel of our old M-50 amplifier. But people wanted the tubes on the front, so we turned the chassis around when we introduced the MkIIs. The MA-2s still have a front panel though.
We still do our anodized finishes at the same place ARC used to do them.
I think the model 8 was a better sounding unit too. The 9 IMO only worked acceptably in Triode mode. Side note: when I was first showing my employees what I wanted for hand-wiring quality, I showed them the insides of a model 8b we were renovating at the time and told them this was the quality of workmanship we had to meet and beat.
|
I forgot the great Leak and Quad amps, low powered but wonderful. |
Ralph, your champagne anodizing is a MUCH better version than that of ARC. When I got my SP-3, D51, and D75, I was disappointed with the rather garish gold anodizing of their faceplates. Your champagne is more like that of Conrad-Johnson, understated and elegant, very tasteful. And your build quality is far higher than that of ARC, imo. Plus, no circuit boards! |
The inside of Ralph’s amps are pieces of art! Love the layout and skilled construction. |
@atmasphere To the third, a correct answer requires a correct question- the **response** is not raised; the **pre-emphasis** is, and by about 5 db (so the simple answer is thus 'no'). Remember that the curve is just that and is the amalgam of three timing constants. The range of frequencies to which you refer is the area where the response is relatively flat compared to the rest of the curve. It isn't flat because of the ordinates of the timing constants. So its a gentle curve in that range, and the **pre-emphasis** is 'lifted' by about 5 db over that range, not 12.
There are plenty of images on the web. Fig. 2 of the article at the link below is pretty good- you can see the 5 db or so rise in the pre-emphasis in the range of frequencies about which you asked:
We are talking about just two time constants that are at 500 HZ and 2200 Hz. Surely you know that time constants or frequency corners are defined points where the response is 3 db up or down. Since these are first order they eventually (within the audio range) come to be 6 dB. Thus 6 db rise for the first pole and 6 db loss for the second. 6+6 = 12 dB. This is also confirmed by graph you presented from Stereophile. I am simply trying to tell readers that the high end of a disc is cut 12 dB higher above 2200 Hz. Also that the RIAA playback curve they are accustomed to seeing is for a magnetic (velocity) cartridge and the one for a displacement (strain gague) would be only 12 dB top to bottom as apposed to 40 dB. How can you disagree? |
I forgot the great Leak and Quad amps, low powered but wonderful .
They are indeed. The Leak is a rather standard circuit as I recall. However the QUAD 2 has a very interesting circuit. Unique, thoughtful. I sell more of my 35 watt than my 100 watt amplifier. I also get calls for the little EM7 2.5 watt single ended. How can some do with so little power and others "think" they need all those watts? Truth is nobody measures and the majority guessing are way off. These older amps may be low powered in todays numbers but adequate for listening with reasonable speakers. I saw a 190# SS amp for sale here for $40,000. Now who needs that. |
Roger, there is a current Audiogon thread inquiring about the Dynaco ST70. I posted that for the same money one could get a used RM-10, a far better amp. If you want, find the thread and let the op know why ;-) . |
can you find the thread. I had no luck. |
|
|
How can some do with so little power and others "think" they need all those watts? Truth is nobody measures and the majority guessing are way off. Some of us both measure and calculate ;-) My 36w PL5 calculates to 93db¹ @ 3.75m with 86db/V²/m speakers. Add in a couple of db crest factor, and I'm outta gas @ 95db peak. Confirmed with scope and RTA. To reach 105db, I'd need 323w for unclipped peaks. 150wpc SS only gets me to ~100db peak because SS clips less gracefully. A 9 watter requires a listening distance of about 1.9m or 92db/V/m loudspeakers. ¹ = room effects ignored ² = 2.83V |
This is also confirmed by graph you presented from Stereophile.
I am simply trying to tell readers that the high end of a disc is cut 12 dB higher above 2200 Hz. Also that the RIAA playback curve they are accustomed to seeing is for a magnetic (velocity) cartridge and the one for a displacement (strain gague) would be only 12 dB top to bottom as apposed to 40 dB.
How can you disagree?
The graph shown by Stereophile is correct, and it shows **less than 6dB rise between 500 and 2KHz**. Again, the gradients of the graph are 5dB increments, and at 500Hz is only about 2.5dB down from where it is at 1KHz. And a similar amount up at 2KHz. With regards to the rise above 2Khz; at 20Khz its about a 15db rise; but the disk cutter is not limited to 20KHz so the pre-emphasis goes all the way up until reaching the cutter amp’s bandwidth limit. To prevent phase shift, the playback should have bandwidth with attendant EQ to at least the same frequency and most of them do (and most LOMC cartridges have the bandwidth as well). However, constant amplitude transducers like the strain gauge and ceramic cartridges aren’t properly equalized as you point out unless some additional EQ is applied. This isn’t a problem for most people though (because strain gauge cartridges are so rare that statistically they don’t exist, having so many zeros as significant digits that any numbers to the right of the decimal point that aren’t zeros are probably a figment of someone’s imagination; my respects to Douglas Adams) so really isn’t much of a concern, except for the new strain gauge cartridge guy that thinks he’s going to make a million bucks on his new entry into the market. If he doesn’t have that equalizer sorted out, he likely won’t be selling his cartridge for very long. Panasonic seemed to be the only one that really had this sorted to any degree and even they bailed on it. |
Thanks Al and Chris. I really gotta learn how to do that ;-) . |
