estuart said:
I'm sorry. Of course, I mean her eyes.
It's her eyes that you were looking at?????????????
Continuing the impedance matching bla..bla..bla.. Mosfets are inherently will have high gate input impedance (BJTs tends to have low base input impedance), while having low output impedance.
I like John Curl's approach of putting mosfet for VAS and for predriver. This is elegant approach, without any additional components, the impedance matching requirement (between critical stages within a power amp) is fullfilled.
Or even better, doing things like Borbely and Nelson Pass (look at A75) to put all transistors Fet and Mosfets. This approach will automaticly have better source-receiver impedance matching in any stage within a power amp without any addittional components (compared if all transistors are BJT's, need extra component or extra thinking to make impedance matching in some points)
I like John Curl's approach of putting mosfet for VAS and for predriver. This is elegant approach, without any additional components, the impedance matching requirement (between critical stages within a power amp) is fullfilled.
Or even better, doing things like Borbely and Nelson Pass (look at A75) to put all transistors Fet and Mosfets. This approach will automaticly have better source-receiver impedance matching in any stage within a power amp without any addittional components (compared if all transistors are BJT's, need extra component or extra thinking to make impedance matching in some points)
lumanauw said:
G’day lumanauw
I don’t think that you can use the term “impedance matching” here. The term has connotations to RF design and means something completely different to what you are describing.
MOSFET’s also have large inter-electrode capacitances. A common source MOSFET VAS will perform poorly unless driven from a relatively low impedance source. This compromises the gain of the input long tail pair, as here, for high gain, a high load impedance is required. Because of this, a MOSFET VAS will generally require a BJT emitter follower (or similar) to buffer it from the LTP if such a compromise is to be avoided. In my opinion, in this regard, there really isn’t any benefit to using a MOSFET VAS at all.
Also, compared to BJT’s, MOSFET’s have low transconductance. This translates to lower VAS gain. Also, a typical BJT VAS emitter degenerated to give a net gm still a few times larger than a typical MOSFET VAS will have not only more gain but significant better linearity as well.
Some people prefer MOSFET VA stages because of their speed and achievable slew rate. In my opinion though a BJT VAS can still be made to perform with a perfectly ample slew rate, and besides speed, a MOSFET VAS looses out in just about every other department.
FET’s and MOSFET’s are also harder to design with than BJT’s, due to their large parameter spreads. This means either component matching or circuit trimming.
Despite all the shortcomings of a MOSFET VAS, some people still insist on using them because they insist that they sound better. That hasn’t been my experience though 🙄
Sure, if you’re willing to accept all of the icky design and performance compromises that necessarily come with such a simplistic design philosophy………
Cheers,
Glen
Hi, Glen, Darkfenriz,
Yes, wrong usage of word. I shouldn't write "impedance matching", but "impedance bridging". I don't know what to call my bla..bla..bla..before I see the wikipedia 😀
Glen, besides the low GM of mosfets and wide spread specs, what are the disadvantages for using mosfet in power amps? What about running the mosfets hot, is there any advantage?
Yes, wrong usage of word. I shouldn't write "impedance matching", but "impedance bridging". I don't know what to call my bla..bla..bla..before I see the wikipedia 😀
Glen, besides the low GM of mosfets and wide spread specs, what are the disadvantages for using mosfet in power amps? What about running the mosfets hot, is there any advantage?
I have just read what it says.darkfenriz said:
Confusing 😱 😕
I thought encyclopedias were supposed to clarify.
WikiP could do with either re-writing that page or removing it completely.
john curl said:
Hi John,
I guess that means that your favored BJT output stages are grainy 🙂.
Bob
AndrewT said:
Andrew,
What's confusing? That link article you pointed to seems clear and correct to me. I hadn't heard the term before, but the dsecription makes sense.
Jan Didden
Hi Janneman,
you must have a better understanding than me.
I have trouble with the analogies.
Zload >> Zsource and says "Essentially no power is transferred". Then he quotes a speaker, talking about priorty of control, but here is an example where I thought power was delivered to the load even though it appears to fit that impedance matching requirement. Note that I said impedance matching not impedance equalising (where very limited total power demands that maximum power be transferred). That brings up another anomally that leads to confusion, when is power transfer of low signal levels important? he quotes no examples (later he looks at transmission lines and misses the opportunity to show the need for equalising at rf signal detection).
Then he mentions electric motors (maximising power transfer) but as far as I know they too are driven from low impedance sources and do not require impedance matching/equalising implied in his statement.
He goes on to talk about "load impedances must be matched to prevent reflections of the signal at the ends of the line from causing reflections and echoes". But, here he is clearly talking about impedance equalising.
I think the writer has got himself confused and as a result is finding it difficult to explain the differences correctly between his different examples in a manner that is understandable to a reader who needs the information contained in this Wiki page.
I would be tempted to say that his use of the word "bridging" (I too am unfamiliar with this usage) is an excellent example of impedance matching and it would be a good example to show the 10:1 ratio often adopted in audio line level signal transfer.
you must have a better understanding than me.
I have trouble with the analogies.
Zload >> Zsource and says "Essentially no power is transferred". Then he quotes a speaker, talking about priorty of control, but here is an example where I thought power was delivered to the load even though it appears to fit that impedance matching requirement. Note that I said impedance matching not impedance equalising (where very limited total power demands that maximum power be transferred). That brings up another anomally that leads to confusion, when is power transfer of low signal levels important? he quotes no examples (later he looks at transmission lines and misses the opportunity to show the need for equalising at rf signal detection).
Then he mentions electric motors (maximising power transfer) but as far as I know they too are driven from low impedance sources and do not require impedance matching/equalising implied in his statement.
He goes on to talk about "load impedances must be matched to prevent reflections of the signal at the ends of the line from causing reflections and echoes". But, here he is clearly talking about impedance equalising.
I think the writer has got himself confused and as a result is finding it difficult to explain the differences correctly between his different examples in a manner that is understandable to a reader who needs the information contained in this Wiki page.
I would be tempted to say that his use of the word "bridging" (I too am unfamiliar with this usage) is an excellent example of impedance matching and it would be a good example to show the 10:1 ratio often adopted in audio line level signal transfer.
Bob, I only use bipolar transistors as output devices, You use them throughout in your circuits. Please don't try to make frivilous comparisons between your circuits and mine.
It's true that I would prefer fets as output devices, however I don't find them as practical in my relatively powerful power amps. You know: 100-800W. If I made a 50W amp or maybe a bridged output amp, I would use all fets in my designs and drop the bipolar transistors.
I don't often contribute here, and when I do, I hope to inject an informed opinion. I don't appreciate backtalk or misconstruing of what I am trying to say.
It's true that I would prefer fets as output devices, however I don't find them as practical in my relatively powerful power amps. You know: 100-800W. If I made a 50W amp or maybe a bridged output amp, I would use all fets in my designs and drop the bipolar transistors.
I don't often contribute here, and when I do, I hope to inject an informed opinion. I don't appreciate backtalk or misconstruing of what I am trying to say.
john curl said:
Mr.Curl,
Could you please tell us why did you prefer FETs for a 50W amp and not the powerful Bipolars....Is there any thing special with Fets in 50W power amplification level...
regards,
Kanwar
Hi, Mr.Curl,
I think what Mr.Cordell saying in post#927 is a friendly joke/chat. Moreover, it has a smile at the end of the post 😀
We (at least I) always waiting for breadcrumbs from you 😀
I think what Mr.Cordell saying in post#927 is a friendly joke/chat. Moreover, it has a smile at the end of the post 😀
We (at least I) always waiting for breadcrumbs from you 😀
I had bad luck 20 years ago trying to make a really reliable 100W/channel all fet power amp. I just could not keep the IRF140 and 9140 devices completely safe from breakdown if the output was accidently shorted.
I have found, in my experience, that bipolar transistors could be used as output devices, and they were easier to parallel for more peak current, and heat spread throughout the heatsink than most power fets.
My conversations with Chas Hansen of Ayre, confirmed most of my findings.
However, I think that fets potentially have a smoother transfer function and therefore tend to generate less higher order order distortion, which I think is most important. I therefore use fets where I can, even though I have perhaps 100 or more of different types of bipolar transistors in my lab to experiment with, if I chose to.
Charles Hansen has done some new research with bipolar transistors, and perhaps he will open a new interest in me to use them again.
I have found, in my experience, that bipolar transistors could be used as output devices, and they were easier to parallel for more peak current, and heat spread throughout the heatsink than most power fets.
My conversations with Chas Hansen of Ayre, confirmed most of my findings.
However, I think that fets potentially have a smoother transfer function and therefore tend to generate less higher order order distortion, which I think is most important. I therefore use fets where I can, even though I have perhaps 100 or more of different types of bipolar transistors in my lab to experiment with, if I chose to.
Charles Hansen has done some new research with bipolar transistors, and perhaps he will open a new interest in me to use them again.
john curl said:
Thanks Mr.Curl ,
I have though succeded in preventing any type of breakdown in Vertical mosfets whether its a continuous short circuit....
I have seen your Parasound amp in which you have used lateral Fet as drivers for the output devices which indeed were bipolars large paralleled pairs...I like the setup very much...[And an INTER-M amp M-1000 which was exact copy of your design]
Mr.Hansen once used vertical mosfets but then moved on to laterals in his designs..
regards,
Kanwar
AndrewT said:
Andrew,
I understood his speaker case as pointing out that you want very low Zout of the power amp to control the speaker, not to maximize power transfer. You maximize power transfer by keeping the power dissipation in the source as low as possible wrt the power dissipated in the load. That is not the same as 'maximise control' which requires lowest Zout. So in this respect I can see his statement that both line-level and speaker level signal 'bridging' is the same in concept.
I am not sure what you mean exactly with 'impedance equalisation', but for me impedance matching is needed to minimize reflections etc in hf transmission-type connections, i.e. you make source- and load impedance (and transmission line characteristic impedance) the same. I think we agree here.
Jan Didden
john curl said:
Hi John,
A little touchy today?
You continue with your misconception that big amplifiers can't be made with MOSFET output stages; that is not an informed opinion. Maybe YOU can't do it, but plenty of others can and have.
Bob
Bob, my experience was 20 years ago, and I have not tried to make an all fet power amp since then. However, I don't see why the IRF devices are really much better than 20 years ago, and I can make successful audio products with bipolar output devices. Perhaps, you should try to remove some of the bipolar devices and replace them with FETs from your power amp, to strive for all FET perfection, which I presume that Erno is recommending.
john curl said:
Hi John,
If I based my design philosophy on what I failed at 20 years ago I would not be a very good engineer, now would I?
You're right, the IRFP240/9240 is probably not much different than what was available 20 years ago (I don't remember when it was introduced). But that's not the point. The point has to do with how the device is applied. If I made a JC-1 (call it a BC-1 🙂) out of nine pairs of IRFP240 and IRFP9240 in the output stage, I'm sure it would be just as robust as your design, it would probably have better bias stability, and the cost of the output devices would be about half. I'm probably better at building MOSFET output stages than you are, and you are almost certainly better than I am at building BJT output stages than I am. That's just the way life is. We all make design choices that we are comforatble with. There is nothing wrong with you using BJTs and there is nothing wrong with me using MOSFETs.
And we have all blown up a lot of devices along the way - that is how we learn. In fact, I'm an equal-opportunity device destroyer - I've destroyed vacuum tubes, BJTs and MOSFETs.
Cheers,
Bob
fab said:
If you are willing to pay the price Borbely (http://www.borbelyaudio.com/audiophile_components.asp) can match them for you. Each type (N and P) had respectively the same lot number and they were matched from N to P to within 10% VGS from what I recall. I use 5 pairs in my power amp.
Source resistor is not necessary at all when using matched lateral mosfet devices.
Also, have in mind that Lateral mosfet are a lot more expensive than vertical ones...maybe because of low volume...
I stand corrected by Erno today and this is his comment on my post:
"...I can match P-P and N-N, but NOT N-P! The ones you are getting (or got, 🙂) were matched at say 0.9V for the N and 1.1V for the P.
Out of several hundred you might get an N-P match, but I guess even very large number of devices would not yield more than a couple of kits..."
Taming The Vertical Limits...
Hello Bob,
Verticals Verticals Verticals....These mosfets seems to be a nightmare for some designers, but why??😉
Your Favourite Mosfet pair IRFP240/9240 is now residing with Vishay at his new residence:
http://www.vishay.com/irf-products/
Cheers,
Kanwar
Hello Bob,
Verticals Verticals Verticals....These mosfets seems to be a nightmare for some designers, but why??😉
Your Favourite Mosfet pair IRFP240/9240 is now residing with Vishay at his new residence:
http://www.vishay.com/irf-products/
Cheers,
Kanwar