Hi Mark,
Which book? I know of no public errata page(s) for the power amp book but I have some notes written down.
Same goes for the new book titled “Audio circuits and systems” which Imo should have its own thread so that the two do not get mixed up. I was hoping that someone else would do that but if not, I have no problem doing it. If I start the new thread I get to own editing the first post allowing me to continue adding new errata assuming that’s the way it works 🙂
Rick
Which book? I know of no public errata page(s) for the power amp book but I have some notes written down.
Same goes for the new book titled “Audio circuits and systems” which Imo should have its own thread so that the two do not get mixed up. I was hoping that someone else would do that but if not, I have no problem doing it. If I start the new thread I get to own editing the first post allowing me to continue adding new errata assuming that’s the way it works 🙂
Rick
Last edited:
Designing Audio Circuits and Systems, the new one, sorry I hit this thread searching for "Bob Cordell errata"
Hi Rick,
Starting a new thread for the new book on audio circuits and systems is a great idea and I would appreciate you doing that. I have been remiss in the past in putting up a section on my web page noting errata, partly because I was working pretty intensely on writing the book. This is also true for the power amp book. I'm now in the process of updating numerous parts of my web site, so I'll try to put in a list of errata as they become know, and any that you have would be much appreciated to be posted here so I can include them.
Cheers,
Bob
Starting a new thread for the new book on audio circuits and systems is a great idea and I would appreciate you doing that. I have been remiss in the past in putting up a section on my web page noting errata, partly because I was working pretty intensely on writing the book. This is also true for the power amp book. I'm now in the process of updating numerous parts of my web site, so I'll try to put in a list of errata as they become know, and any that you have would be much appreciated to be posted here so I can include them.
Cheers,
Bob
Thanks, Rick.
I should also mention to the good folks here that Rick helped greatly in my writing of the new audio circuits and systems book. Rick read every chapter of the draft and provided numerous suggestions and caught many needed corrections. Many thanks, Rick.
Cheers,
Bob
I should also mention to the good folks here that Rick helped greatly in my writing of the new audio circuits and systems book. Rick read every chapter of the draft and provided numerous suggestions and caught many needed corrections. Many thanks, Rick.
Cheers,
Bob
I should also mention that I have been somewhat AWOL from the Forum for quite awhile, and for that I apologize. For some reason I had not been getting email notifications of many Forum posts. I had actually been wondering if Forum activity on the threads that I participate in had fallen off. I'll go back to the beginning of June and look at some of the posts to which I should reply and replay to many of them.
Cheers,
Bob
Cheers,
Bob
I'm sorry to hear of your unfortunate experience with the physical quality of the book, and am impressed and flattered that you paid so much extra for the hardcopy edition. I hope that you will enjoy the content in spite of the concerns about the physical construction of the book. I'll relay your concerns to my contact at Routledge.Since I don't own the paperback edition, I cannot really tell. Routledge is using some sort of ultrathin and flexible glue for the spine of paperbacks, at least that is the one they used in Douglas Self's book, and it also appears to be the same one used in Bob's hardback version. Sorry for bringing back Doug's book again and again, but it is the most recent Routledge book I have bought previous to Bob's book. Anyway, back to the glue. It gives the impression of being a good thing because it allows for the paperback book to open flat on a table, but it also looks extremely brittle and appears like it will crack in no time, it even looks like some pages are about to fall out. The spine looks cracked and several thick lines started appearing all along the spine.
Regarding the hardback version of Bob's book, which I do have, it is also not very good quality (the book itself, not the content). If you try to open the book flat on a table, you'll start hearing cracking noises and the spine starts ungluing from the hardcover. It is a scam (by Routledge, not Bob), really, so I have to treat it with upmost care and try not to open the book flat, but rather open it partially on my left hand. As an irrelevant note: I am not sure why Routledge hardback versions have some generic blue cover rather than the cover used for the paperback. I also think they might be using Amazon's printing services in TX.
Sorry if all of this sounds obnoxious or picky to you. I own an extremely large book collection (great to have, painful to move) and I have spent a considerable amount of my hard-earned money on books. I am tired of paying for books with a price tag over the hundred dollar mark that end up cracking, ungluing, or that are just subpar in terms of construction quality. I believe publishers are getting more and more greedy each passing day, and that they are taking far more advantage of their authors and buyers than they used to. Although I also must admit that they are still a necessary evil, since I always frown upon any book on Amazon labeled as "Independently Published".
Cheers,
Bob
Thank you, Bob. The book seems majestic so far, but the build quality is the same-old-same-old for me, as I am getting used to this treatment from publishers. Thanks!
I recommend the following medium-Idss LSK489 model:What are people using for an LSK489 SPICE model? The above was from 2018 is the standard library best?
* LSK489B IDSS = 8.64mA Rds = 147
.MODEL LSK489B NJF (LEVEL=1 BETA=19.0E-4 VTO=-2.68 LAMBDA=1E-3
- IS=3E-15 N=1 RD=75 RS=65 CGD=3E-12 CGS=8E-12 PB=0.8 FC=0.5
- KF=1.5e-18 AF=1 XTI=0)
*
The model in the Linear Systems online models may be wrong.
I use the LSK489 and LSK389 dual JFETs a lot. Numerous example designs in my new book use them. They are excellent parts. For example, I show a moving coil preamp that uses the two halves of an LSK389 as the JFET input transistor connected in parallel (since the two sides are well-matches) to achieve 1 nV/rt Hz.
Cheers,
Bob
Thank you for pointing this out. You are correct, I had forgotten that the Tiger output stage had gain. It seemed like a good idea at the time in order to reduce the needed VAS signal swing, Nowadays, a good VAS has very low distortion even with a signal swing near the rails. As you point out, there can be problems like the one you pointed out in output stages with gain.@Bob Cordell
I've been digging into the second edition of your book and I notice on pg. 293 Section 13.2
that you mention the first use of a CFP output stage being by Dan Meyer in the Universal
Tiger 1970. That amp was not a traditional CFP but rather it had gain. The NPN device made
to act like a PNP in a quasi-comp output stage is CFP and those were common in the mid to
late 1960s. Any reference to the Universal Tiger should go in your section 13.3 on "CFP Output
Stages With Gain" in my opinion. It is clear that the Bryston borrowed a lot from the Tigersaurus,
yes with many improvements.
I was the one to discover the reverse Vbe voltage problem in 2011 as documented in this post:
And in both the Tiger and the Bryston.
https://www.diyaudio.com/community/...r-improved-and-simulation.93637/#post-2585028
After a lot of suggestions regarding how to fix the reverse Vbe problem I came up with the
"Basel diode" lol in 2020, two diodes completely fix the problem as shown here:
https://www.diyaudio.com/community/...oved-and-simulation.93637/page-5#post-6041249
I think that it is important to mention the reverse Vbe issue and a good correction.
In your list of references #5 is to the Tigersaurus and you refer to it under "CFP Triple With Gain"
on page 297. That amp is not a triple, but it does have stacked outputs to tolerate the 75V rails.
The Tiger .01 had an output stage triple.
The first and only quasi-complementary output stage amplifier I did was circa 1970 when I built a clone of the Harman-Cardon Citation 12. However, I am now in the process of redesigning a Phase Linear 400 whose output stage is a quasi-complementary Triple. It looks to be a well-deigned quasi-comp output stage, but it is pretty much biased into class B. I'll keep the quasi-comp output stage and bias it into class AB and see if the crossover distortion is sufficiently low, maybe with a few tweaks. I'll build a completely new IPS/VAS. I'll also add a much-needed speaker protection circuit (some people referred to the company name as "Flame Linear").
Cheers,
Bob
Thanks Bob, and by the way we're discussing the PL700 here starting at post #22:
https://www.diyaudio.com/community/threads/power-amp-stability-analysis-in-older-amps.414946/page-2
I guessed that it would be an SOA nightmare and that was confirmed. MOSFETS should fix that,
not that anyone will ever try the idea.
https://www.diyaudio.com/community/threads/power-amp-stability-analysis-in-older-amps.414946/page-2
I guessed that it would be an SOA nightmare and that was confirmed. MOSFETS should fix that,
not that anyone will ever try the idea.
Funny you should mention the PL700 and MOSFETs. I've been doing some work on a PL700 amplifier I acquired about 10 years ago. It came with no output transistors or circuit board or reservoir caps. I acquired new new caps and have done a little on it over the years. My plan has been to do a completely new design with vertical MOSFETs and a good IPS/VAS with a JFET input stage and DC servo. I also plan to add error correction to it.
In many ways, it will be a monsterized version of the MOSFET amplifier with Error correction I built circa 1984 and which is described in the JAES paper on my web site.
It is a big, long-term job and it will be a fun challenge. As I'm sure you know, that amp is a beast. At no-load, its rails are at about 105 V. SOA was definitely an issue for that design, even with 5 output pairs. To address that, the design used fairly aggressive V-I limiting. There were some who referred to the PL700 as a "tweeter eater" because of the behavior of the V-I limiter under certain protection conditions.
Thanks for bringing that PL700 thread to my attention; I'll have to take a look at it.
Cheers,
Bob
In many ways, it will be a monsterized version of the MOSFET amplifier with Error correction I built circa 1984 and which is described in the JAES paper on my web site.
It is a big, long-term job and it will be a fun challenge. As I'm sure you know, that amp is a beast. At no-load, its rails are at about 105 V. SOA was definitely an issue for that design, even with 5 output pairs. To address that, the design used fairly aggressive V-I limiting. There were some who referred to the PL700 as a "tweeter eater" because of the behavior of the V-I limiter under certain protection conditions.
Thanks for bringing that PL700 thread to my attention; I'll have to take a look at it.
Cheers,
Bob
Hi Bob,I recommend the following medium-Idss LSK489 model:
* LSK489B IDSS = 8.64mA Rds = 147
.MODEL LSK489B NJF (LEVEL=1 BETA=19.0E-4 VTO=-2.68 LAMBDA=1E-3
*
- IS=3E-15 N=1 RD=75 RS=65 CGD=3E-12 CGS=8E-12 PB=0.8 FC=0.5
- KF=1.5e-18 AF=1 XTI=0)
*
The model in the Linear Systems online models may be wrong.
I use the LSK489 and LSK389 dual JFETs a lot. Numerous example designs in my new book use them. They are excellent parts. For example, I show a moving coil preamp that uses the two halves of an LSK389 as the JFET input transistor connected in parallel (since the two sides are well-matches) to achieve 1 nV/rt Hz.
Cheers,
Bob
and what about LSJ689 spice model? Do you have a good model?
I use one model from Linear Systems but I don´t know if works ok.
I came to the conclusion that to keep the PL700 with BJT outputs it needs a series output
stage, and there's room for 6 pairs/ch on the heatsink if the drivers are moved to the board.
Bryston used one for a long time even when BJTs could take the voltage, probably for better SOA.
Anyway, MOSFETs are also a good idea, double die for high current and lower losses.
I looked up the DH500 and was surprised that it only uses 3 pairs/ch of the 8A, 125W parts.
I own one, have driven it hard into 3 ohm loads and it has never failed. Still, I'd rather see
3 pairs of the double die version.
stage, and there's room for 6 pairs/ch on the heatsink if the drivers are moved to the board.
Bryston used one for a long time even when BJTs could take the voltage, probably for better SOA.
Anyway, MOSFETs are also a good idea, double die for high current and lower losses.
I looked up the DH500 and was surprised that it only uses 3 pairs/ch of the 8A, 125W parts.
I own one, have driven it hard into 3 ohm loads and it has never failed. Still, I'd rather see
3 pairs of the double die version.
The Medium - IDss LJ689 = 6.8 mA Model I use is the following:Hi Bob,
and what about LSJ689 spice model? Do you have a good model?
I use one model from Linear Systems but I don´t know if works ok.
*
- .MODEL LSJ689B PJF (LEVEL=1 BETA=30E-4 VTO=-1.75 LAMBDA=2E-3
- + IS=4.5E-16 N=1 RD=99 RS=37 CGD=6E-12 CGS=11E-12 PB=0.25 MJ=0.3 FC=0.5
- + KF=2E-18 AF=1 XTI=0)
*
Cheers
Bob
The MOSFETs I'll be using for the PL700 are verical MOSFETs (IRFP240 and IRFP9240), si I don't need double die, as with laterals. I'll be using the plastic devices, unlike the TO-3 parts used in the original PL700, so I will be mounting them on a 1/8-inch aluminum heat spreader fastened to the back of each heat sink location on the inside of each heat sink unit. Each of the 4 heat spreaders is basically the full height and width of each heat sink fin. I think that, given the large area of the interface, I'll get sufficient heat transfer through the stell chassis from the heat spreaders to the heat sink fins, using a thin layer of thermal grease at the interfaces..
I'l bias each of the 5 pairs (or is it six - I have not looked in awhile) at 150 mA, for a total idle bias of 750 mA. This means that each channel will have 750 mA of quiescent bias, making for a good-sized class Q region. At 200 V real-to-rail, each channel will dissipate about 150 watts at idle.
As you probably know, Rick Savas and I have designed and implemented what we call the DH-220C redesign of the Hafler DH220. It uses all of the original lateral MOSFETs and has a completely new circuit board (which is available). A separate small circuit board fits in the heat sink area where the output transistors used to be be wired point-to point. The design also works with the DH200, PA230 and DH500 (with some revised circuit boards for the latter. The amplifiers work quite well. It is handy to make a new amplifier design when you already have all of the basic infrastructure.
The new design uses a full-complementary JFET input stage with LSK489 and LSJ689 dual monolithic devices.
Cheers,
Bob
I'l bias each of the 5 pairs (or is it six - I have not looked in awhile) at 150 mA, for a total idle bias of 750 mA. This means that each channel will have 750 mA of quiescent bias, making for a good-sized class Q region. At 200 V real-to-rail, each channel will dissipate about 150 watts at idle.
As you probably know, Rick Savas and I have designed and implemented what we call the DH-220C redesign of the Hafler DH220. It uses all of the original lateral MOSFETs and has a completely new circuit board (which is available). A separate small circuit board fits in the heat sink area where the output transistors used to be be wired point-to point. The design also works with the DH200, PA230 and DH500 (with some revised circuit boards for the latter. The amplifiers work quite well. It is handy to make a new amplifier design when you already have all of the basic infrastructure.
The new design uses a full-complementary JFET input stage with LSK489 and LSJ689 dual monolithic devices.
Cheers,
Bob
Thanks, Bob.The Medium - IDss LJ689 = 6.8 mA Model I use is the following:
*
*
- .MODEL LSJ689B PJF (LEVEL=1 BETA=30E-4 VTO=-1.75 LAMBDA=2E-3
- + IS=4.5E-16 N=1 RD=99 RS=37 CGD=6E-12 CGS=11E-12 PB=0.25 MJ=0.3 FC=0.5
- + KF=2E-18 AF=1 XTI=0)
*
Cheers
Bob
I use this ( Idss = 14.6mA )
.MODEL JLSJ689_4 PJF (LEVEL=1 BETA=28E-4 VTO=-2.75 LAMBDA=2E-3
- IS=4.5E-16 N= 1 RD=73 RS=35 CGD=6E-12 CGS=11E-12 PB=0.25 FC=0.5
- KF=2E-18 AF=1 XTI=0)
Both models have different capacitances compared to values shown in the datasheet, when VDS = 10V. ( less than datatasheet ).
So, I did put two capacitors in the circuit simulation. In my case, VDS is constante because the jFets are in cascode configuration.
I believe that way the combination ( junction capacitance model and added capacitors ) values, are near to real ones.
eD
Last edited:
Curious as to why you chose verticals, IRFP240 and IRFP9240. I've known about these for a long time,The MOSFETs I'll be using for the PL700 are verical MOSFETs (IRFP240 and IRFP9240), si I don't need double die, as with laterals. I'll be using the plastic devices, unlike the TO-3 parts used in the original PL700, so I will be mounting them on a 1/8-inch aluminum heat spreader fastened to the back of each heat sink location on the inside of each heat sink unit. Each of the 4 heat spreaders is basically the full height and width of each heat sink fin. I think that, given the large area of the interface, I'll get sufficient heat transfer through the stell chassis from the heat spreaders to the heat sink fins, using a thin layer of thermal grease at the interfaces..
I'l bias each of the 5 pairs (or is it six - I have not looked in awhile) at 150 mA, for a total idle bias of 750 mA. This means that each channel will have 750 mA of quiescent bias, making for a good-sized class Q region. At 200 V real-to-rail, each channel will dissipate about 150 watts at idle.
As you probably know, Rick Savas and I have designed and implemented what we call the DH-220C redesign of the Hafler DH220. It uses all of the original lateral MOSFETs and has a completely new circuit board (which is available). A separate small circuit board fits in the heat sink area where the output transistors used to be be wired point-to point. The design also works with the DH200, PA230 and DH500 (with some revised circuit boards for the latter. The amplifiers work quite well. It is handy to make a new amplifier design when you already have all of the basic infrastructure.
The new design uses a full-complementary JFET input stage with LSK489 and LSJ689 dual monolithic devices.
Cheers,
Bob
since Nelson's article, but never looked into the specs, are they trench type or? I am aware that they
are popular, but why them over the Exicons?
I'm sure Bob will respond, but when I used the old Hitachi 2SK somethings I was not overly impressed with the performance. The typical frequency response was little better or the same as the epi bJT's (around 2MHz).
I've not used Exicons as I consider them to be largely the same tech. Vertical types like IRFP240 are faster. However, I originally considered the 9240 not to be an exact complement when I first looked at these -at the time "new device type". It seemed that the next higher rated PMOS device matched the NMOS better, until the improved gain NMOS devices appeared which disrupted the matching somewhat again. Though with several in parallel I suspect there will be enough amperage not to be too concerned about matching. At the time I was considering them, I was thinking of only a single pair in lower power amps. and attempting to get the best match.
I've not used Exicons as I consider them to be largely the same tech. Vertical types like IRFP240 are faster. However, I originally considered the 9240 not to be an exact complement when I first looked at these -at the time "new device type". It seemed that the next higher rated PMOS device matched the NMOS better, until the improved gain NMOS devices appeared which disrupted the matching somewhat again. Though with several in parallel I suspect there will be enough amperage not to be too concerned about matching. At the time I was considering them, I was thinking of only a single pair in lower power amps. and attempting to get the best match.
From what I remember of the test reports on all of the MOSFET Haflers was that they
could produce full power at 20KHz, even above, maybe 50KHz without any problem
at all. No cross conduction problems, and that's what I see as the big advantage.
Also, I believe as a result of no SOA issues the DH500 only requires 3 pairs of 8A 125W
output devices, and is rated to drive 2 ohm loads, try that with BJTs. They have
significant advantages as I see it.
could produce full power at 20KHz, even above, maybe 50KHz without any problem
at all. No cross conduction problems, and that's what I see as the big advantage.
Also, I believe as a result of no SOA issues the DH500 only requires 3 pairs of 8A 125W
output devices, and is rated to drive 2 ohm loads, try that with BJTs. They have
significant advantages as I see it.
- Home
- Amplifiers
- Solid State
- Bob Cordell's Power amplifier book