| JanDH |
As a novice I have looked at a lot of amplifier designs and decide to go for the Aleph-X. I’am going to build a three way active cross-over and of course three times an amplifier per channel. I’am thinking of : 20 Watts for the tweeter, 30 watts for midrange and 150 watts for the woofer. I noticed that multiple IRFP044N are placed parallel in order to obtain a certain capacity.
- Well why don’t I see the IRFP064N or IRFP1405 applied.
- Why are there in the power stage relatively many resistors used. Would eg the MP930 series from Caddock not do very well. High power and low induction.
- Does it make any sense to separate the power for drivers and power stage? (Regulated power for driver?)
- I noticed a relative low rail voltage. Why not going for a higher supply in combination with eg IRFP1405
Pls any advice is very welcome in order to make the right decisions for the project I have in mind.
Thanks!! |
|
|
| Mark A. Gulbrandsen |
" Well why don’t I see the IRFP064N or IRFP1405 applied."
I suppose you could if the capacitance, transconductance and other parameters are similar to IRFP240.
" Why are there in the power stage relatively many resistors used. Would eg the MP930 series from Caddock not do very well. High power and low induction."
Many have used the Caddock and other similar type resistors in the Aleph amplifiers. I doubt that you would hear much of a difference between them and high grade 1% wire wounds though. The Caddocks also may need to be heat sink mounted in order for them to be able to operate at max dissipation. That is somewhat of a disadvantage along with would amount to a very marginal gain in sound quality if any at all. They are also very expensive.
"Does it make any sense to separate the power for drivers and power stage? (Regulated power for driver?)"
Actually there is a constant current source for the input stage, and in these Alephs ther is only an input and an output stage anyway. There is not really a driver stage in there per say.
" I noticed a relative low rail voltage. Why not going for a higher supply in combination with eg IRFP1405"
An X amplifiers rail voltage requirement is dependent on a number of things including output stage bias, output power requirements, and power required into low impedances, say 4 ohms. It would be nice to hear from Grey Rollins on this one
Also keep in mind that all the semiconductor devices need to be hyper matched, or even better, matched in place in the operating amplifier circuit you intend to use. This is necessary to achieve the lowest possible DC offset at the output terminals. So be ready and set up for doing that!!
Mark |
|
|
| JanDH |
Hi Mark,
Thanks for your reply.
My comment starts with '-'
" Well why don’t I see the IRFP064N or IRFP1405 applied."
I suppose you could if the capacitance, transconductance and other parameters are similar to IRFP240.
- Yes I think you right however devices parrallel also double the capacity, at least I learned so.
" Why are there in the power stage relatively many resistors used. Would eg the MP930 series from Caddock not do very well. High power and low induction."
Many have used the Caddock and other similar type resistors in the Aleph amplifiers. I doubt that you would hear much of a difference between them and high grade 1% wire wounds though. The Caddocks also may need to be heat sink mounted in order for them to be able to operate at max dissipation. That is somewhat of a disadvantage along with would amount to a very marginal gain in sound quality if any at all. They are also very expensive.
- Oke but we are working on these projects because we want it all. Some folks think that parallel resistors (or evices) create e 'foggy sound'. The prices from mouser.com are acceptable.
"Does it make any sense to separate the power for drivers and power stage? (Regulated power for driver?)"
Actually there is a constant current source for the input stage, and in these Alephs ther is only an input and an output stage anyway. There is not really a driver stage in there per say.
- Almost every amplifier has a current source. People like http://www.borbelyaudio.com/ and http://www.lcaudio.com/ go for this theory.
" I noticed a relative low rail voltage. Why not going for a higher supply in combination with eg IRFP1405"
An X amplifiers rail voltage requirement is dependent on a number of things including output stage bias, output power requirements, and power required into low impedances, say 4 ohms. It would be nice to hear from Grey Rollins on this one
Also keep in mind that all the semiconductor devices need to be hyper matched, or even better, matched in place in the operating amplifier circuit you intend to use. This is necessary to achieve the lowest possible DC offset at the output terminals. So be ready and set up for doing that!!
- Yes as I noticed from the excel configuration sheets. It is difficult to order all info from this site and filter only this information that is important.
However I get the strange feeling that this amplifier is a potential dangour for my valuable speakers isn't it.
Regards,
Jan |
|
|
| HBarske |
As stated in some other threads around here: if possible, avoid using TO220 devices as output transistors. Though the IRF1405 looks impressive from the datasheet, you will run into problems if you have to get rid of serious amounts of heat. I have cooked lots of them in high power dc motor driving applications at only 10 watts of power losses.
And, those extremely high power FETs indeed have an enourmous amount of input capacitance, you will need serious driving current to feed them. |
|
|
| jwb |
The IRFP044, 140, 240, etc. series of TO-247 transistors really are the best. They can dissipate quite large heat loads without much trouble.
I think you'll find that if you are using parallel FETs in the output stage, you probably don't need to use huge power resistors. In my Aleph-X I was able to use BC 1W metal film resistors for every place in the power stage. They are far less bulky than 3W types and non-inductive. |
|
|
| Nelson Pass |
Please note that both the Aleph and Aleph X amplifiers
require that the negative rail of the front end requires
the same voltage connection as the output stage.
The only thing you can gain with a separate + supply for the
front end is perhaps a bit less noise, but even then you can
achieve this with some decoupling to the supply rail which
powers the current source for the input diff pair. |
|
|
| GRollins |
There's nothing sacred about the '044. If you've got some other device you'd like to try, feel free to do so. Just keep an eye on the SOA for whatever you decide to put in.
I use Caddock resistors in mine. I've got/had several versions of the Aleph-X drawn up (found some schematics that seem to have survived the recent hard disk adventure--thank goodness for floppies). As I said in the Aleph-X thread, the circuit I run isn't quite the same as the one I published. Given the high cost of Caddock and other boutique resistors, I felt that the circuit would reach a wider audience if it didn't call for comparatively rare, mega-buck parts. It's the values that are important. If you want to use Vishay, Caddock, Mills, etc. and your wallet is thick enough, then go for it. Consider heat sinks for the Caddocks, as they're only good for a couple of watts without them. The internal solder connections come unstuck if you overdo it.
Regulate the entire rail if you really want to max the thing out. I'll eventually go that route, but am not doing so yet. The regulator circuit is ready and tested and I've even got the parts, what I lack is the time to do a circuit board.
I'm not sure why people are so startled by the low rails; it's a bridged amp, and low rails/high current go with the territory. If you want to use higher rails, you'll end up with a higher powered amp. No problem. But...you'll have to increase the bias in order to keep from current limiting into lower impedances, and that implies lots and lots of heat. You'll need big heat sinks. Or you could consider going water cooled. Broadly speaking, the rails determine the output wattage, and the bias current determines how low an impedance you can deliver that wattage to. If you're not careful, you'll end up with an amp that can deliver 500W...into a 100 ohm load. Clearly not a practical amp for general use.
Grey |
|
|
| macka2 |
Grey & Nelson,
At the moment I am looking to scale several X Alephs to bi amp and eventually quad amp
I have done a search but not found anything specific regards low supply rail limitations.
I recall Nelson at one point saying idealy the X Aleph is best with 25 volts. Is that per rail or rail to rail.
Just curious.
Ian |
|
|
| jwb |
| Mine are working with ± 14V DC rails, and I find it perfectly fine. Of course it will depend on the load you want to drive. If you are going to biamp or even quadamp a speaker, you can obviously afford to build smaller, less powerful amps. |
|
|
| GRollins |
Ian,
At the moment, I've got an Aleph-X running at roughly 20V rails (a little over 60W output) and a bodacious amount of bias current. No real game plan there, it's just that I had a pair of power transformers on hand that seemed lonely.
Regarding distortion:
Some preach the Gospel of Bias Current as though it's the only thing going. Ain't so. Rail voltage is important, too. Somewhere in the original Aleph-X thread, I posted some numbers relating rail voltage to distortion. No, I don't remember where, but it's near the back of the thread, so start at the end and work forwards. The upshot is that increasing rail voltage decreases distortion up (or down, depending on your point of view) to a certain point. Obviously, you'll never get to zero. Where's optimal? Well, that's a complex question. You've got to balance rail voltage (higher is better), against bias current (higher is better), and heat dissipation (less is better), the capability of your chosen power transformer, heatsinks, the thickness of your wallet, etc. etc. etc.
It's enough to drive a man to drink.
Hmmm...interesting idea, that one.
There's no one right answer, just as there's not really a wrong one. The 15V rails I showed in the original schematic are there because I developed the Mini-A first, and it ended up having 15V rails because my target power was 10W. The Aleph-X is essentially two Mini-As back to back. There's nothing sacred about 15V rails. I ran up the Aleph-X in its posted version just as a demonstration of concept, not something to be slavishly adhered to. Feel free to adapt the critter to whatever rail voltages/currents you have available. As long as you attend to a few part values, it'll take pretty much whatever you throw at it.
Note that I don't take credit for the thing. Nelson just left a few building blocks on the table and I got to playing with them, seeing whether thay would fit together. They did. But they're his building blocks. He was gracious enough to let me play with them.
Grey |
|
|
| Nelson Pass |
The reason that I don't like to go much below 25 volts (+ and -
each) is that Mosfets really like to see a few volts across them,
even at peaks. The issue is the capacitance of the Mosfet which
increases non-linearly as the voltage decreases across the
device, and is a priimary form of distortion at high frequencies.
Just take a look at the capacitance curves.
So I like to keep 5 volts or so left for may gain devices at full
peak output, which means I can get +-20 volts clean off any
output stage, or twice that in the case of an X circuit. |
|
|
| macka2 |
Thankyou Nelson and Grey for your responses.
I will work through this information and refer to the original thread.
Ian |
|
|
| macka2 |
Quote:
The reason that I don't like to go much below 25 volts (+ and -
each) is that Mosfets really like to see a few volts across them,
even at peaks. The issue is the capacitance of the Mosfet which
increases non-linearly as the voltage decreases across the
device, and is a priimary form of distortion at high frequencies.
Can you explain the non linear capacitance with voltage and its effect on the circuit in this situation?
What I mean is as we increase the supply rails what tends to happen that improves this aspect of performance and the result of adding devices to share the power dissipation?
At the moment my Drain / Source voltage is about 17 volts.
Aside from the lower capacitance, the curve for the 240 appears to flatten at a lower Vds.
Ian |
|
|
| Blues |
| quote: | Originally posted by Nelson Pass
The reason that I don't like to go much below 25 volts (+ and -
each) is that Mosfets really like to see a few volts across them,
even at peaks. The issue is the capacitance of the Mosfet which
increases non-linearly as the voltage decreases across the
device, and is a priimary form of distortion at high frequencies.
Just take a look at the capacitance curves.
So I like to keep 5 volts or so left for may gain devices at full
peak output, which means I can get +-20 volts clean off any
output stage, or twice that in the case of an X circuit. |
I think I heard the effect of this in my zv4 when i played with the R4 pot. 20-22V across Q1 took the life out of piano notes...less weight and the timbre was not right. The sound is livelier and sweeter (IMO) at about 25V...went as far as 30V but after several hours smoked Q1 which in turn took down Q2 and Q5. At 25V the positive cycle will clip/distort earlier than the negative cycle (about 20V)...but at my normal listening level the measured AC voltage at the spkr terminals was about 3.25Vrms (~4.6Vpk)...the first watt is what really matters! |
|
|
| macka2 |
Yeah well those curves are for a reason.
I think this is what makes building up these designs so interesting...you come across a problem, investigate whats happening and learn something.
I'm sure these a lot more to scaling these amps than keying numbers in to a speadsheet.(although its a very useful tool)
Exactly what impact raising the voltage 10 volts apart from pumping up the watts will be interesting with the IRFP044's. I have a tube of 25 of these suckers so I hope it works.
I will try this tonight and post the results.
Ian |
|
|
| Nelson Pass |
As long as you can keep 5 volts or so minimum across the
Mosfets over the operating cyle, your larger problem will
tend to be the bias current. Keep it high ;) |
|
|
| macka2 |
Hmm,
Those IRP044's a tough, but I might use 2 pair per side and be damned. That will be 4800 pf Gate capacitance in theory (or the equivalent of say 4 IRFP240's) so it will either work or it won't.
As an amusing side note, I have found the Olympic broadcasts a good test signal at least for the Aussie viewers, the applause is much like averaged white noise and its easy to differential spectral balance.
:cool:
Ian |
|
|
| promitheus |
hi,
I can find the IRFP044N alot easier than the IRFP044. They are almost the same but the "N" type have a lower input capacitance.
The IRFP044 have 2500pF and the IRFP044N have only 1500pF.
Wouldn´t they be a better choice? |
|
|
| macka |
I have not done the test yet due to other priorities but I note from Nelson's excellent how to article on bias/voltage comparions the problems of Gate Capacitance and selection of mosfets are well illustrated.
For my purposes at least the 044 is less than suitable as I have no intention of biasing the device at say 2 amps and 25 volts for credible performance, particularly for high frequencies where I need it.
I hope I am missing something here but looking at 150-200 watts dissipation for a tweeter amp in a quad amped system with tweeter 105 db seems a little bit silly and perhaps the X Aleph or even is less appropriate for this application.
Maybe a batch of IFR610 would be the way to go?? I think the Rushmore uses scaled down Alephs, the total power dissipation being around 300 watts total per box.
I will probably re install the JLH which uses 80 watts and is to my ear as good or better in some respects in low power applications.
macka |
|
|
| grataku |
The N is a newer device and It would probably be better.
on the subject of bias: one can always choose to run a higher voltage/current ratio while using easy loads like broadband high efficiency driver without X-overs, no?
I am still waiting for Nelson's big claw... |
|
|
| keypunch |
Nelson,
I am not an electronics engineer, not even as hobby. I just know enough to be "dangerous" ;) .
I have couple questions if I may re your post:
| quote: | Originally posted by Nelson Pass
The reason that I don't like to go much below 25 volts (+ and -
each) is that Mosfets really like to see a few volts across them,
even at peaks. The issue is the capacitance of the Mosfet which
increases non-linearly as the voltage decreases across the
device, and is a priimary form of distortion at high frequencies.
Just take a look at the capacitance curves.
So I like to keep 5 volts or so left for may gain devices at full
peak output, which means I can get +-20 volts clean off any
output stage, or twice that in the case of an X circuit. |
1) Is this mostly related to Class A amps?
2) If no to (1) above, what threads, forms, documents, et al could you suggest one read, even if target audience is electronics engineers, that I could read and see what I can digest out of?
3) Which specs are you referring to as capacitance that is listed on the MOSFET specifications?
I was asking as I have a couple of 225VA 15VAC based toroidal transformers I was thinking I use for small 15W amps.
Regards,
John L. Males
Willowdale, Ontario
Canada
12 December 2004 19:03 |
|
|
| GRollins |
John,
1-That's basic MOSFET behavior, and will be pretty much the case regardless of operating class of the amplifier.
2-There's nowhere to go to get a "one stop shop" answer for these sorts of things. The de facto standard for books is Horowitz & Hill, The Art of Electronics. There are a zillion other places you can go to get small pieces of the puzzle, but that's as good a reference as you're going to find for anything relating to audio. There have been books written with audio in mind, but they're either hopelessly narrow-minded or too specialized to be a good general reference text for audio use. Face it, we're a very, very small niche market and there's not much money to be made on audio design books.
3-Start with www.irf.com and look up the spec sheets for whatever device you're interested in. They're available as PDF files, so you can print 'em off and wallpaper your kitchen with them. That's not to say that your interior designer will think much of your choice in wallpaper, but it's the latest thing in electro-geek.
Are the transformers 15-0-15 or just a single 15V secondary? You can get things to work either way, but you have to hook them up differently. A 15V secondary will give you roughly 20V rails. That's good for a nearly 20W amp as long as you don't get radical with the bias current. As a matter of fact, I've got a Mini-A in the living room right now that runs under almost exactly those conditions.
Grey |
|
|
| keypunch |
Hi Grey,
(Hope Grey does not refer to grey hair with all the IT challenges I have seen you comment on odd time from IT real world ;) )
Thanks for your reply and answers.
1) Thanks for answer. I had asked as I am not sure other Class A amp heat issues aside, if I have much power to draw from. I am limited to a sort of 15A circuit can use, unless run couple extensions in from kitch to living room :(( So I am thinking maybe I build a few to have and enjoy, and also have some low wattage Class AB amps as well, ergo the question.
2) Sorry I was not implying one stop shopping. I am already trying to piece and organize the pieces of information I have read. It will burn up a few DLT tapes and more than few CDs ;) I had feeling this is collection process and cross referencing of raw information. That was the contect of the question, sorry to have not been articulate that was assumption. Expect this collection process will consume about 200 GB of space very easy if not 400GB. Hours to do so, well you have some clue I am sure. Distill the information ... umm ... Time to Distill in hours = size of data in bytes ** 10!
3) Already have lots of the IFP, and many other datasheets in PDF on my system. Lord only knows how many times I have looks at a number of these datasheets. I was asking if I could be enlightened as to what spec item on datasheet "as capacitance". I know tehre are lots of capacitance specs, just not sure which one you may be referring to or what otehr specs one refers to to calculate this "capacitance" you have refered to. My cats are not fussy about the interior design of place. They in fact have their own opinions of interior design anyway ;)
As to the 225VA 15VAC based toroidal transformers I have, each has a primary 117VAC, and 4 seperate (not centered tapped), yes 4, 15VAC secondaries. I have sent in an inquiry to find out what is the rating of each secondary. The rating is for 7.5 amps. I do not know if that is 7.5 amps for ceach secondary or 7.5/4=1.875A a secondary or a "floating" rating in that any sum of the secondaries must not exceed 7.5A. 20W of power is just fine for me based on 15VAC supply. I have a spreadsheet to calculate VAC to rail to rms/peak -+V and A, which suggested about 16W of power. 16W is ok by me as well. The spreadsheet is not slanted to the Aleph design, but at least I get an basic ballpark good enough for my selection needs.
Oh, another question. What is the AC at mains draw for your Mini-Aleph? If you know please advise what transformer, amp bias, amp output driver using and number of Mini-Alephs connected to supply?
Regards,
John L. Males
Willowdale, Ontario
Canada
12 December 2004 22:57 |
|
|
| GRollins |
The capacitance in question is the Gate capacitance.
Egad, I use junk-box transformers to build Mini-As. There's no telling what the things are rated at. I just pick it up, heft it, scowl at it...you know, the kinda things that you do when you're doing Science. If it's not too rusty and I want the shelf space where it was sitting for something else, I toss it in a circuit. Did this recently for a headphone amp. The rail voltage was too high for what I intended, so I grubbed around and found some parts and built a CRCRC power supply to rub off a couple of the unnecessary volts. Silver lining to the cloud was that the rail was absolutely quiet after all that massaging--still a clean, flat line on the smallest resolution on my scope.
Grey |
|
|
| wuffwaff |
Promitheus,
the 044N has a lower possible heat dissipation I think 100 instead of 150 watts and a much higher Rth. This means that you must use a lot more fets to keep the junction temp low enough.
William |
|
|
| jacco vermeulen |
To me it would seem that if the IRFP240 or IRFP044 are hard to get replacing it by IRFP044N is not the obvious choice.
e.g. :
Three IRFP630 in parallel have a lower Cgs and Cds than a single IRFP044.
Rjs = 0.73 for the IRFP630's and Rjs = 1.54 for the IRFP044N.
A negative difference is Rds which is 0.13 for the 630's and 0.020 Ohm for the 044N.
At 8 amps of bias that reads as a bit more than 1 volt drop instead of 0.16 Volt for the 044N, not that impressive.
As cooling is the prime directive for these components in Aleph circuits cheap 630's that can be had anywhere for a dime are a better choice than 044N's.
Even if they have a TO220 package, three midgets can conquer one giant.
Reichelt sells the 240, even at a very attractive price, i miss the point. |
|
|
| buglehead |
I have recently finished an Aleph 4 using 8 irfp260s per channel. I did this just to get it finished at the least cost since I already had a bunch of 260s. To me, the amp sounds great, but I now need to decide if I should replace them because of the capacitance issue. Does anyone know from experience what effect on the sound the high capacitance has at high frequency? My amp goes to prettty high frequencies, and I do notice a very faint sssssss sound at the top. This sound is similar but more faint than in many commercial Mosfet amps I have heard that claim very low distortion. Does the capacitance cause roll-off on the high end? In all respects, the amp sounds spectacular, so I may need some convincing that I need to swap out these IRFP260s.
BTW, I want to thank Nelson Pass (The One And Only) and many others on this forum, like GRollins, for making my project possible. |
|
|
| keypunch |
| quote: | Originally posted by buglehead
I have recently finished an Aleph 4 using 8 irfp260s per channel. I did this just to get it finished at the least cost since I already had a bunch of 260s. To me, the amp sounds great, but I now need to decide if I should replace them because of the capacitance issue. Does anyone know from experience what effect on the sound the high capacitance has at high frequency? My amp goes to prettty high frequencies, and I do notice a very faint sssssss sound at the top. This sound is similar but more faint than in many commercial Mosfet amps I have heard that claim very low distortion. Does the capacitance cause roll-off on the high end? In all respects, the amp sounds spectacular, so I may need some convincing that I need to swap out these IRFP260s.
BTW, I want to thank Nelson Pass (The One And Only) and many others on this forum, like GRollins, for making my project possible. |
buglehead,
Interesting observations. I have been labouring over the various spec of many of the popular IRF's used by many of the projects here. Part of my motivation is to have sense if there be any corelation of performance the ears hear to specs. I have been doing some non-engineering evaluation of the specs and some "interesting" math on the specs.
Assuming the issues you have noted are based on the output driver IRFP260, if you have IRFP260N it would be interesting to know what you think of the quality from IFRP260N. Based on my number crunching you might want to try a IRFP264N or IRFP250N as well or instead of the IRFP260N. These suggestions are based purely on non-engineering based math I have done on the specs of the some IRF Mosfets. That said, others far more qualified will be able to suggest any points about bias, part value changes or supply changes specific devices require within the design. It goes without saying being able to keep as many of the parameters the same if one does compare different output devices, this will enable one a more objective manner to evaluate any sound quality differences. I realize what is being evaluated in terms of sound quality is subjective, but at moment you are the same subjective evaluator and I assume that other than the changes you make to the Aleph-X all the otehr items in the audio food chair (speakers, pre-apms, room, material, wires, etc) remain the same.
Regards,
John L. Males
Willowdale, Ontario
Canada
13 December 2004 16:24 |
|
|
| jacco vermeulen |
I don't have the datasheet for the IRFP260s.
IRFP260 has a tremendous high Cgs, more than 5000pF.
Using 8 means Cgs = 40nF.
What is Cgs for the IRFP260s ?
I can imagine that with such high Cgs you'll have slewrate problems at high frequencies, even if the real Cgs for the given output may be significantly smaller.
Your story confirms NP's statement that SR figures are not all that important. |
|
|
| buglehead |
Thanks for the replies.
I should have been more specific about the chip. It is IRFP260N. I'm sorry but IRFP260s does not exist, it is the plural of IRFP260.
Ciss = 4057. I'm using 4 per rail. I finished the amp about 2 months ago and it has settled in pretty well. Good soundstage and imaging. Great bass. I have used massive heatsinks and can bias it up, which I should probably try in any case. It would be a lot of work to swap out output devices, so I would probably try it once only. I think I would rather spend the energy on a pair of Aleph-X. I would use the "proper" output devices, and would then have something to suitable to compare against. |
|
|
| macka |
Nelson Pass has an excellent article on mosfet properties (and matching) at Passdiy.com
macka |
|
|
| GRollins |
The primary effect of higher Gate capacitance is increased distortion at higher frequencies.
Grey |
|
|
|
