aleph 1.2's to AX100
GL thanks, I just found a pile of 1ohm 2 watt metal films, so I guess it must be done and I will replace the 1.5 ohms. I also don't need 200watts but this is the output that the trannie will give me. I have to leave on a business trip to India (16hour plane ride from NYC) and this will give me time to ponder my design decisions. dave
GL thanks, I just found a pile of 1ohm 2 watt metal films, so I guess it must be done and I will replace the 1.5 ohms. I also don't need 200watts but this is the output that the trannie will give me. I have to leave on a business trip to India (16hour plane ride from NYC) and this will give me time to ponder my design decisions. dave
gl said:
In reality, this is not a deal of knowing or not the simulator, which is the most commen one : PSpice!
This is just a deal of understanding the circuit: that's all.
Hope GRollin or somebodyelse could post it giving me so, some good insight.
I don't understand if it's possible that the bottom part draws more current than the upper parts..and if not.....try to understand why and how can be fixed.
Anyways that you for taking the time...at least...to read the post!
Best.
Stefanoo said:
and where is that current difference supposed to go..........?
you are talking about DC or AC current ?
I can bet that source resistors are mismatched
Stefanoo said:
I hate to burst your bubble, but there is actually quite frequently a major difference between the simulator and the real world....hence the comment made by gl, since he is experienced enough to know that you NEED to know the simulator to know why the simulation acts up. His comment is what you would get from an experienced electronics guy whom have seen such differences as the above mentioned only too often.......
Magura 🙂
I can't read the circuit values on the schematic you attached but more likely the CCS feedback resistor values on each half are not set for a 50% AC Current Gain. Try adjusting values up/down for "supersymmetry" current 🙂
Magura said:
As i said: this simulator is one of the most commen in use...PSpice.
I didn't know that each simulator program could give you, possibily, a different result...and you have to interpretate this result based on the kwnoledge that you have on the simulator itselfs.
This doesn't really make much sense to me.....but if you say it....as i consider myself a newbie, i will take this observation in serious consideration.
Stefanoo said:
Numeric mathematics 101 example:
X1 + 1.001 * X2 = 2
1.001 *X1 + X2 = 2
= > => X1 ~ 1, X2 ~ 1
Enter a small alteration :
X1 + 1.001 * X2 = 2.1
1.001 * X1 + X2 = 2
=> => X1 ~ 49 , X2 ~ 51
A 5% change in the input parameter leads to a 5000% change in the outcome, and X1/X2 are also no longer equal.
Simulation programs are extremely nice, they can save a lot of time and do calculations that would take years or a lifetime to do manually.
But simulations are always approximations, and even the input is not exact.
The outcome of a simulation is by definition never completely accurate, and should not be taken for granted.
There are however methods to calculate the maximum difference between the exact result and the approximation.
Simulation software do an endless number of matrix calculations to produce the results, i'm sure a lot of folks overhere do Spice simulations without even knowing exactly what a matrix or a Taylor series is.
For the larger part, because the software is free and the user can simulate a circuit within a day's time without having a clue on how the circuit operates.
Understanding what something does is often more important than the number on the paper it produces.
There's not a calculator or computer that can solve every imaginable calculation.
No1 rule in engineering : KISS
A traditional engineering approach, start with basic approximations ; sizes, weight, power requirement.
Step 2 : use these variables as the input for iterations.
An exotic heatsink in a 300 x 200mm size that can do 0.1C/W is a Seifert SK271.
You are starting off with the biggest Aleph-X amp ever, judging by the numbers you've posted.
Right from the start, you're picking the most efficient and expensive heatsinks.
And from the start, you are trusting the outcome of a simulation.
Then let me give you another example.
There are 3-D software packages that pretty accurately calculate the lift an airplane wing generates.
But the actual wing may not be as smooth, thereby lowering the lift number.
Ice deposits on the wing not only increase the weight of the airplane, but also have a negative effect on the lift factor.
There are some other variables to consider which influence the lift number.
Calculating the simulated lift number to the last digit is a waste of time, provided the max offset percentage is determined.
At the end of the line, a percentage is added to make sure the primary design demands are met.
In commercial production, missing the set requirements of the contract during the final test run before the owner takes possession means having to pay a huge penalty fine.
Your 0.1C/W heatsinks may or may not meet the specs, depending on a couple of factors.
A lot of class A amps i see on pictures posted here have great heatsinks, but often the amp case rests on small feet.
I'm a drag queen power amp type myself, out of the ordinary heatsinks such as the SK271 better stand on 2" high heels if the spec needs to be reached.
Steve Slater was kind enough to send me a big box of Fat Boy rubber feet for the junk an idiot like me assembles.
Assuming a heatsink will do what's printed on the paper is taking another step towards Armageddon.
The one who desires to build a supercar usually has been on the track for at least a decade.
There are 3-D software packages that pretty accurately calculate the lift an airplane wing generates.
But the actual wing may not be as smooth, thereby lowering the lift number.
Ice deposits on the wing not only increase the weight of the airplane, but also have a negative effect on the lift factor.
There are some other variables to consider which influence the lift number.
Calculating the simulated lift number to the last digit is a waste of time, provided the max offset percentage is determined.
At the end of the line, a percentage is added to make sure the primary design demands are met.
In commercial production, missing the set requirements of the contract during the final test run before the owner takes possession means having to pay a huge penalty fine.
Your 0.1C/W heatsinks may or may not meet the specs, depending on a couple of factors.
A lot of class A amps i see on pictures posted here have great heatsinks, but often the amp case rests on small feet.
I'm a drag queen power amp type myself, out of the ordinary heatsinks such as the SK271 better stand on 2" high heels if the spec needs to be reached.
Steve Slater was kind enough to send me a big box of Fat Boy rubber feet for the junk an idiot like me assembles.
Assuming a heatsink will do what's printed on the paper is taking another step towards Armageddon.
The one who desires to build a supercar usually has been on the track for at least a decade.
jacco vermeulen said:
I did some tests how far the heatsinks need to be from the floor and the like to offer their max.. 100mm (4") is just enough to face the law of diminishing returns 😉
Magura 🙂
jacco vermeulen said:
Believe or not . . .
I love using aluminium chassis instead of normal steel box. You know? The aluminium plate is very bad heat insulator between the inner chassis temprature and the ambient. Therefore, I could save air flow holes which I should have in case of the normal steel box, and could have dirst-tight.
Pass DIY Addict
Joined 2000
Paid Member
The Seifert heat sinks are VERY nice sinks, but I also think you'll end up short in the heatsink department. The Seifert KL271 (300mm wide) specs that I have are as follows (black anodize, natural convection):
150mm length = 0.22c/w
200mm length = 0.18c/w ($62 each in 2005)
250mm length = 0.16c/w
300mm length = 0.145c/w
Then, practice (not simulation) indicates that you really need to de-rate these specs by about 25% to get realistic performance figures. Thus, if you apply 150w to power transistors mounted on a 200mm length of KL271, theory indicates approx 27c temp rise. For the several amps I've built, there are inefficiencies that make up the 25% loss. Thus a 0.18c/w heat sink really behaves more like a 0.23 or 0.24c/w heatsink. Thus, this same 150w applied to a 200mm length of 271 will more likely result in a temp rise of about 35c.
The bottom line (at least to me) is that if you want to build a single chassis to dissipate nearly 600w, you should be thinking of using 4 KL-271s per chassis. This will likely provide a 35c temp rise over ambient. If you want to keep temp rise to more like 25c that Nelson recommends, you are looking at 6 KL-271s per 600w chassis.
The Aleph-X amps that I am working on are planned for 100w/8ohms and dissipate 300w on heatsinking of 0.06c/w per channel.
Of course, I could be wrong.
Eric
Hi Eric,
How are your Aleph-X's coming along. I love your web site by the way - being an old A40 fan myself. But there hasn't been much activity for a while.
Graeme
How are your Aleph-X's coming along. I love your web site by the way - being an old A40 fan myself. But there hasn't been much activity for a while.
Graeme
Pass DIY Addict
Joined 2000
Paid Member
Hey Graeme,
Unfortunately, my Aleph-X's are still just populated boards, matched fets, and a pile of trannies. The $1000 that I had earmarked for heatsinks, chassis components, and other remaining misc parts just got dumped into my wife's car in the form of a new air conditioning compressor
The air died the day after I began pricing the sinks...
When I finally get my hands on some heatsinks (did I mention the huge landscaping project I'm also working on?) my web page will come alive again with updates based on my progress. I really only need some sinks and rectifiers in order to fire one of the channels up. It seems like forever ago that I was matching my output mosfets!
Then I need to build cabinets and crossovers for my new speaker project...
Someone I know once said: "slow work takes time"
Unfortunately, my Aleph-X's are still just populated boards, matched fets, and a pile of trannies. The $1000 that I had earmarked for heatsinks, chassis components, and other remaining misc parts just got dumped into my wife's car in the form of a new air conditioning compressor
The air died the day after I began pricing the sinks...When I finally get my hands on some heatsinks (did I mention the huge landscaping project I'm also working on?) my web page will come alive again with updates based on my progress. I really only need some sinks and rectifiers in order to fire one of the channels up. It seems like forever ago that I was matching my output mosfets!
Then I need to build cabinets and crossovers for my new speaker project...
Someone I know once said: "slow work takes time"
Eric,
my heart bleeds for you, especially when you're talking airco compressor.
Since the switch from Freon to R134a, the price tag of those things seem to rise by the month.
If the rates at the gas station don't kill you, the aircon does.
Here's the C/W graph of the SK271, but only a clever pony can reach those numbers.
www.digtion-medien.de/seifert/Uploads/seifert_punkte/pdfs/47[0].pdf
my heart bleeds for you, especially when you're talking airco compressor.
Since the switch from Freon to R134a, the price tag of those things seem to rise by the month.
If the rates at the gas station don't kill you, the aircon does.
Here's the C/W graph of the SK271, but only a clever pony can reach those numbers.
www.digtion-medien.de/seifert/Uploads/seifert_punkte/pdfs/47[0].pdf
Pass DIY Addict
Joined 2000
Paid Member
Jacco,
Yeah, and the 134a doesn't work as good as good ole fashioned freon either! Glad I don't have to commute anymore - I used to drive 120 miles a day, now I drive 3. I do know a few people who are still driving those distances are getting killed by gas prices. When I was in Holland two summers ago I noticed the price there was the same as it was in the US - only the unit of measure was different. I remember something like $1.57 per liter when it was $1.56 per gallon in the US.
The numbers for the 271 sinks that I provided earlier are from the distributor that I purchased them from and they seem to be pretty accurate. I don't know where the numbers in that chart from Seifert actually come from - perhaps they were measuring on the moon where its REALLY cold!
You need to be clever indeed!
Eric
Yeah, and the 134a doesn't work as good as good ole fashioned freon either! Glad I don't have to commute anymore - I used to drive 120 miles a day, now I drive 3. I do know a few people who are still driving those distances are getting killed by gas prices. When I was in Holland two summers ago I noticed the price there was the same as it was in the US - only the unit of measure was different. I remember something like $1.57 per liter when it was $1.56 per gallon in the US.
The numbers for the 271 sinks that I provided earlier are from the distributor that I purchased them from and they seem to be pretty accurate. I don't know where the numbers in that chart from Seifert actually come from - perhaps they were measuring on the moon where its REALLY cold!
You need to be clever indeed!
Eric
Magura said:
Just to complete this subject:
I found the reason of why there was more current drawn from the bottom part. (the super symetryc configuration wasn't propertly configured).
Therefore, playing around with the bias of the input differential stage, the mos output bias, the source resistors and the CCS, i was finally able, at least "on the paper", to reach certain wanted specifics for this amp and, at least, they seem to be allright to me.
If i was going to build this circuit thinking that the simulator was wrong, i was gonna build a failure circuit, for sure!
Obviously the vice-versa is not guaranteed as well (now that the simulation results are correct won't necessary imply that the circuit will run for sure).
I just think that, at least, if you are NOT super experienced like me, to not be able to catch at the first look how the things are going on that specific circuit, the simulation will be very helpfull to find potential failures and to understand the main parameters of that particular topology....obviously this can't replace the theorical part (you have to know, at least, the basics of the electronic) and, obviously, the practical part (after the simulation....building parttt yehaaaaa!!! 🙂 ).
😎
Best,
Stefano.
Stefanoo said:
I am an engineer.
My way of engineering approach is first looking at the whole picture (forest), to avoid any global failure or global stupid things. I do not know the simulator very well. Nevertheless, I believe that it would help for me to see the big picture.
The Next step I take is looking into the details (trees) to find out any small mistakes or for trimming off unnecessary parts. I think that here we have many items and details which decide the final quality. On this stage, we need to use the basic knowledge, common sense and experience.
Once I have my own confidence with the big picture, I just spend my most times for the quality in detail.
I think you are doing the same!
Aleph 1.2 to AX100-160
GL and Grey, I am redesigning the Power supply of my Aleph 1.2 to upgrade it to GL design. I found four 600va trannies and I thought I could do the following:
1. Use two per monoblock for a total of 1.2kva
2. I would use one trannie for the positive supply and one for the negative supply.
3. I would connect the centertap of the two transformers AFTER the politically correct hexfred diode bridges. One diode bridge for the positive diode supply and one for the negative. Is this ok in your thinking? Because these transformers can be vertically stacked in the amp I can pick up some valuable real estate to add extra filter caps.
4. This needs some explanation. I was reading an article in The Absolute sound about the BAT VK 600 SE power amp which is rated at 300w/ch @8 ohms and 600 @ 4ohms. The article contains an interview with the designer who states that this amp has only one mosfet differential gain stage and only N channel fets in the output. The amp is a parallel balanced bridge design like NP's XA's but I think it is class AB. This amp uses PS caps connected directly to each of the output devices and this is called the BAT PACK.
Do you think from a perfromance standpoint that putting Electrolytic caps on the output fets raises the performance in any way from just connecting to the large computer grade caps??
Obviously the benefit is that there is a shorter electrical path and perhaps lower output impedance but at the speed of light does this mean anything? It is a lot more work to do.
Please let me know what you think of the center tap for the trannies. I am also considering uisng a IKVA trannie for each channel but they are bigger in diameter and take up more space (I wouldn't be able to fit any bigger computer grade caps in)
but I can reuse the same hexfred diode bridge and I don't have to add another one.
GL and Grey, I am redesigning the Power supply of my Aleph 1.2 to upgrade it to GL design. I found four 600va trannies and I thought I could do the following:
1. Use two per monoblock for a total of 1.2kva
2. I would use one trannie for the positive supply and one for the negative supply.
3. I would connect the centertap of the two transformers AFTER the politically correct hexfred diode bridges. One diode bridge for the positive diode supply and one for the negative. Is this ok in your thinking? Because these transformers can be vertically stacked in the amp I can pick up some valuable real estate to add extra filter caps.
4. This needs some explanation. I was reading an article in The Absolute sound about the BAT VK 600 SE power amp which is rated at 300w/ch @8 ohms and 600 @ 4ohms. The article contains an interview with the designer who states that this amp has only one mosfet differential gain stage and only N channel fets in the output. The amp is a parallel balanced bridge design like NP's XA's but I think it is class AB. This amp uses PS caps connected directly to each of the output devices and this is called the BAT PACK.
Do you think from a perfromance standpoint that putting Electrolytic caps on the output fets raises the performance in any way from just connecting to the large computer grade caps??
Obviously the benefit is that there is a shorter electrical path and perhaps lower output impedance but at the speed of light does this mean anything? It is a lot more work to do.
Please let me know what you think of the center tap for the trannies. I am also considering uisng a IKVA trannie for each channel but they are bigger in diameter and take up more space (I wouldn't be able to fit any bigger computer grade caps in)
but I can reuse the same hexfred diode bridge and I don't have to add another one.
- Status
- Not open for further replies.