Re: Ace
I've always been impressed with the competency and helpfulness of the workers in the Ace stores. Having said that, thanks to my F5 build experience, I think I can navigate around the "nuts and bolts" boxes better than most of the hired help!
Great store. For now, I'd gladly pay a few cents extra for a store that's magnitudes better than Lowes or Home Depot....!
I've always been impressed with the competency and helpfulness of the workers in the Ace stores. Having said that, thanks to my F5 build experience, I think I can navigate around the "nuts and bolts" boxes better than most of the hired help!
Great store. For now, I'd gladly pay a few cents extra for a store that's magnitudes better than Lowes or Home Depot....!
Surplus Parts in Colorado Springs
6L6, Jim.....
I mentioned a few times a Colorado Springs electronics surplus store in our telephone conversations.
My bad. I didn't realize until a few minutes ago they have an eBay store. Here's the website for it: http://stores.ebay.com/OEM-Parts-Inc
Typical caveats apply: I'm not advocating for OEM, nor do I have any ownership in the company--but it IS a nifty place to roam thru, when/if you are in Colorado Springs.....
6L6, Jim.....
I mentioned a few times a Colorado Springs electronics surplus store in our telephone conversations.
My bad. I didn't realize until a few minutes ago they have an eBay store. Here's the website for it: http://stores.ebay.com/OEM-Parts-Inc
Typical caveats apply: I'm not advocating for OEM, nor do I have any ownership in the company--but it IS a nifty place to roam thru, when/if you are in Colorado Springs.....
Let me try to enter that website for OEM, again!
OEM Parts Inc items - Get great deals on Capacitors, Computer Cards Boards Memory items on eBay Stores!
OEM Parts Inc items - Get great deals on Capacitors, Computer Cards Boards Memory items on eBay Stores!
The business of heatsinks and cooling the Mosfets is way over complicated and mistakes unfortunately do happen, despite the best intentions of computer modellers, graphs, etc.
Simple specs give you a 'ballpark' range of options and this is simply that each channel requires the heatsink (plus whatever case metal screwed to it) to get rid of approx 65 watts/ channel and as heatsink specs are in "X" degrees/ watt, multiplying this rate by those 65 watts gives you the resulting temperature rise that added to your ambient room temp, will tell you how hot they will be in use - farly straight forward.
AS Nelson has said on many occassions, about 50 *C (degrees centigrade) is the point that his production amps run because they sound better and will still run continuously for 20 years at this temperature, so this "expert" should be good enough for you, yes?
Much confision about the transistor case temperature and the transistor junction temperature and it goes like this - heatsink temp about 50*C, transistor (FET) case temp with goo and mica about 70*C, and the junction temp inside the transistor will be about 90*C, that is still well below the "safe" long term temperature of 100*C for those fets.
About your heatsinks, get the specs from the manufacturer and stick them back on eBay as they'll be perfect for someone building one of the higher power Aleph's, etc.
I'm guessing that as your temp rise is only about 15*C above room (25*C?) to give you that "less than 40*C", they'll have a rate below 0.2*C/W (including the reduction for the lower operating temperature - the rate, or thermal efficiency, is measured at 80*C for Conrads, and this is then derated as the sinks dont get rid of as much heat at the lower operating temps.
I don't know if you have done your homework with the First watt F5 article from Nelson but from memory, you need about 4.2volts across the parallel parts of the R3, pots and thermistor + series resistor that will give you an approx 0.5.. across the power resistor between Mfet and rails - same thing for the lower R4 ... set.
Now, silly question, but did ALL the resistors (and thermistor) MEASURE OKAY, including the feedback and the power resistors?
Are those input jfet matched and in the range of about 7mA?
Nothing like doing your own homework, Woody - "experts" are .....
CAM,
I have found that transformer size of 300VA/channel, or a 600 VA with seperate secondaries, makes a quite noticeable difference, and using discrete diodes bridges, not the block bridges (unless they're those IXYs ones), makes a big difference.
As you have access to lab gear, have a look at what the transformer is supplying to the diodes, and what is passed onto the amps power rails, after the power supply caps - need to use high speed of the CRO to see the big nasty turn-on/off diode spikes and the resultant transformer ringing - unfortunately, it's always there! Requires snubber cct to get rid of it.
All the info is in the main thread but buried from awhile back and not so easy to find.
Simple specs give you a 'ballpark' range of options and this is simply that each channel requires the heatsink (plus whatever case metal screwed to it) to get rid of approx 65 watts/ channel and as heatsink specs are in "X" degrees/ watt, multiplying this rate by those 65 watts gives you the resulting temperature rise that added to your ambient room temp, will tell you how hot they will be in use - farly straight forward.
AS Nelson has said on many occassions, about 50 *C (degrees centigrade) is the point that his production amps run because they sound better and will still run continuously for 20 years at this temperature, so this "expert" should be good enough for you, yes?
Much confision about the transistor case temperature and the transistor junction temperature and it goes like this - heatsink temp about 50*C, transistor (FET) case temp with goo and mica about 70*C, and the junction temp inside the transistor will be about 90*C, that is still well below the "safe" long term temperature of 100*C for those fets.
About your heatsinks, get the specs from the manufacturer and stick them back on eBay as they'll be perfect for someone building one of the higher power Aleph's, etc.
I'm guessing that as your temp rise is only about 15*C above room (25*C?) to give you that "less than 40*C", they'll have a rate below 0.2*C/W (including the reduction for the lower operating temperature - the rate, or thermal efficiency, is measured at 80*C for Conrads, and this is then derated as the sinks dont get rid of as much heat at the lower operating temps.
I don't know if you have done your homework with the First watt F5 article from Nelson but from memory, you need about 4.2volts across the parallel parts of the R3, pots and thermistor + series resistor that will give you an approx 0.5.. across the power resistor between Mfet and rails - same thing for the lower R4 ... set.
Now, silly question, but did ALL the resistors (and thermistor) MEASURE OKAY, including the feedback and the power resistors?
Are those input jfet matched and in the range of about 7mA?
Nothing like doing your own homework, Woody - "experts" are .....
CAM,
I have found that transformer size of 300VA/channel, or a 600 VA with seperate secondaries, makes a quite noticeable difference, and using discrete diodes bridges, not the block bridges (unless they're those IXYs ones), makes a big difference.
As you have access to lab gear, have a look at what the transformer is supplying to the diodes, and what is passed onto the amps power rails, after the power supply caps - need to use high speed of the CRO to see the big nasty turn-on/off diode spikes and the resultant transformer ringing - unfortunately, it's always there! Requires snubber cct to get rid of it.
All the info is in the main thread but buried from awhile back and not so easy to find.
I didn't know one could use too much heatsink? I was under the impression that the good sound came from running the MOSFET in its most linear region and not necessarily at a certain temperature.
Peter Daniel has said he heard no difference between two mono block F5s and one stereo F5 all with 300VA transformers. BTW does anyone know where to get IXYS bridges for cheap money?
Update -
I have a number of 2.2k resistors in my box, so I installed another in series with R3 and R4, (Which are the resistors in parallel with the pots) bringing the total resistance to about 4.5k
I have this complete on one channel, it has biased up properly (cold) It is currently heating up. I will check it in a half an hour and see how it's going.
I have a number of 2.2k resistors in my box, so I installed another in series with R3 and R4, (Which are the resistors in parallel with the pots) bringing the total resistance to about 4.5k
I have this complete on one channel, it has biased up properly (cold) It is currently heating up. I will check it in a half an hour and see how it's going.
I used to live in the Springs and when I returned to visit, I couldn't find OEM's current location. I frequented both of the "abandoned-grocery-store strip mall locations." (can't remember street names) The later, about a year ago. Do they still have a store one can walk into and browse or is it all on line?
"Good Sound" (??) comes from a balance of compromises, and the most restrictive is, naturally, the cost of the project.
Some people do feel uncomfortable running their electronics at a higher temp, but it can be useful to gain better results and save some $s at same time. For example, if you are using the "diyAudio" pcbs, if you make up a small bracket that reaches from the heatsink to go between the 2 input jfets and heat them to the higher temperature, they do sound better (IMO, I add!), and this goes against the idea of keeping transistors cool as possible - simple things like this can be "something for next to nothing" bonus.
Don't know why Peter didn't get different results using single and double transformers - quite possibly he uses best parts available and at this level, there is nothing to gain from using the extra VAs - also, he uses high efficiency drivers and possibly doesn't run the amp very hard - can't say. For me, it was obvious, but I'm a bit of a nut about the sound of the supplies!
Out here in Oz, we use 240volt at 50Hz and most of newer Torroids are pretty skimpy when it comes to the cheaper transformers and if "special build" or EI is off the budget, upsizing the Tranxformer is a reasonable compromise.
Instead of the IXYS, use discrete diodes - I genrally use the Isolated tab Phillips BYW series but the 8ETH06s are possible cleaner - cheap ones like the MUR series are also quite okay - Shottkys, unlike Mosfets, DO work best when as cool as possible so need small sinks for them to keep the temp rise as low as possible - don't connect to the chassis/case as it usually gets well above room temp.
6L6,
Is you amp developing approx 4.2 volts across the // combination of R3 (+ new 2k2R), P1 and the THI+ 2k2 ? and this is the same for j74 -> R4 combination?
Are the R1 and R2 actually 10 ohms?
Some people do feel uncomfortable running their electronics at a higher temp, but it can be useful to gain better results and save some $s at same time. For example, if you are using the "diyAudio" pcbs, if you make up a small bracket that reaches from the heatsink to go between the 2 input jfets and heat them to the higher temperature, they do sound better (IMO, I add!), and this goes against the idea of keeping transistors cool as possible - simple things like this can be "something for next to nothing" bonus.
Don't know why Peter didn't get different results using single and double transformers - quite possibly he uses best parts available and at this level, there is nothing to gain from using the extra VAs - also, he uses high efficiency drivers and possibly doesn't run the amp very hard - can't say. For me, it was obvious, but I'm a bit of a nut about the sound of the supplies!
Out here in Oz, we use 240volt at 50Hz and most of newer Torroids are pretty skimpy when it comes to the cheaper transformers and if "special build" or EI is off the budget, upsizing the Tranxformer is a reasonable compromise.
Instead of the IXYS, use discrete diodes - I genrally use the Isolated tab Phillips BYW series but the 8ETH06s are possible cleaner - cheap ones like the MUR series are also quite okay - Shottkys, unlike Mosfets, DO work best when as cool as possible so need small sinks for them to keep the temp rise as low as possible - don't connect to the chassis/case as it usually gets well above room temp.
6L6,
Is you amp developing approx 4.2 volts across the // combination of R3 (+ new 2k2R), P1 and the THI+ 2k2 ? and this is the same for j74 -> R4 combination?
Are the R1 and R2 actually 10 ohms?
I've been using F5 with 100dB Fertins, so they were never pushed hard.
Besides, the implementation and the build of the amp is often more critical than the choice of single or double transformers. I had instances with GC amps, where stereo amp sounded better than dual monoblocks, only because changes have been made either to a chassis design or parts choices.
However, for all my reference amps, I always go with dual mono
Besides, the implementation and the build of the amp is often more critical than the choice of single or double transformers. I had instances with GC amps, where stereo amp sounded better than dual monoblocks, only because changes have been made either to a chassis design or parts choices.
However, for all my reference amps, I always go with dual mono
Last edited:
I had posted all the relevant VAC and DC voltages and current draws earlier in another thread. All indications were a power transformer that was sagging significantly, and our lab testing confirmed that. It appears it even fails to meet the company's published specs, and I will probably "debate" that issue with them. In any case, I have the room in my chassis for a 600VA unit, and I think that will be my first "upgrade" to my F5.
Thanks for the posting--good info....!
RKH...... OEM parts is back in their "original location" at 3029 North Hancock Avenue, Colorado Springs. Their phone is (719) 635-0771. For a while, they were in limbo, moving from an old location back to North Hancock (just north of Fillmore). You might have hit them, while they were in limbo. If there is any downside to OEM, I suspect they have not yet fully unboxed everything from the relocation--but they still have a whole bunch of stuff on their shelves!
Ken
One channel (the first to get the resistors changed) is biasing well. .6v bias and 0 offset. I had to bias it to about .55 when cold, and when warming up it settled into perfect.
The other channel has the same resistors changed, but I'm still running out of range on the potentiometers... It's awfully close, .51v and 5mv offset, but that's all it's got. Cold it wouldn't go past .4v or so.
Should I increase the value of the parallel resistor again on the one channel? It would be easiest to add another 2.2k or thereabouts in series with the other 2 resistors, as that would keep from having to de-solder something from the PCB.
The other question is what is the optimum bias? The recommended .6? As hot as you have heatsink? Or is that not correct? Also, what is the downside of running it a bit cooler? (I.E., the .5 that I have now?)
FWIW, the heatsinks (either the one with full bias and the one a little lower) are cool enough to have your hand on indefinitely. It's toasty warm, but not painful at all.
The other channel has the same resistors changed, but I'm still running out of range on the potentiometers... It's awfully close, .51v and 5mv offset, but that's all it's got. Cold it wouldn't go past .4v or so.
Should I increase the value of the parallel resistor again on the one channel? It would be easiest to add another 2.2k or thereabouts in series with the other 2 resistors, as that would keep from having to de-solder something from the PCB.
The other question is what is the optimum bias? The recommended .6? As hot as you have heatsink? Or is that not correct? Also, what is the downside of running it a bit cooler? (I.E., the .5 that I have now?)
FWIW, the heatsinks (either the one with full bias and the one a little lower) are cool enough to have your hand on indefinitely. It's toasty warm, but not painful at all.
Last edited:
so - now you know recipe - increase them further
seems that :
1 . Idss of used Jfets is small ,or
2. Ugs of used mosfets is unusually big , or
3 . both 1 & 2
on all questions regarding bias - that horse is beaten to death
seems that all greedy boyz agree in fact that let them bleed !! approach is better ; my magic number for max dissipation per mosfet is 50W , but only if you're using IRFP150 and similar biggies
for sissy ones - say that 40W is max
seems that :
1 . Idss of used Jfets is small ,or
2. Ugs of used mosfets is unusually big , or
3 . both 1 & 2
on all questions regarding bias - that horse is beaten to death
seems that all greedy boyz agree in fact that let them bleed !! approach is better ; my magic number for max dissipation per mosfet is 50W , but only if you're using IRFP150 and similar biggies
for sissy ones - say that 40W is max
Last edited:
Up your bias....!!
6L6, JIm......thanks for the telephone call this morning. Seems like things are going in the right direction.
I had asked the question of cool MOSFETS, and the possibility to up bias to get into a "sweet spot" for the amp. This was the comment I got back from Melon Head:
"The limiting factor is the number of output devices you have.
40W/mosfet is the absolute most power I would want to put through them continuously. With 24V rails thats 1.66A. As an experment you could run your amp with 1A then have a listen, 1.3A then have a listen, and if your fearless 1.6A and have a listen. If you want to go more extreme you will need to parrallel the output with another pair (2 mosfet pairs).
(I knew I could find his posting!)
6L6, JIm......thanks for the telephone call this morning. Seems like things are going in the right direction.
I had asked the question of cool MOSFETS, and the possibility to up bias to get into a "sweet spot" for the amp. This was the comment I got back from Melon Head:
"The limiting factor is the number of output devices you have.
40W/mosfet is the absolute most power I would want to put through them continuously. With 24V rails thats 1.66A. As an experment you could run your amp with 1A then have a listen, 1.3A then have a listen, and if your fearless 1.6A and have a listen. If you want to go more extreme you will need to parrallel the output with another pair (2 mosfet pairs).
(I knew I could find his posting!)
No, no and no.
You have to treat Single Ended differently from Push Pull.
You also have to look at the difference that peak current and rms current make to your calculations.
This is an F5 thread, so I'll deal with the F5's Push Pull case.
Bias to 1A allows a Maximum ClassA output current of 2Apk.
Maximum ClassA power is given by Ipk^2 * Rload / 2
For 8r0, this comes to 16W (1AIb)
For 1.3A bias the Max ClassA Power is 27.04W
For 1.6A bias the max ClassA Power is 40.96W
If you feed these same currents into a 4r0 load then the respective maximum Powers are all exactly halved.
But you must also check that the amplifier has not clipped the signal.
2Apk into 8r0 is 16Vpk. That is well within the F5 voltage capability when on +-25Vdc supply rails.
2.6Apk into 8r0 is 20.8Vpk. That is probably right on the nail for maximum unclipped signal.
3.2Apk into 8r0 is 25.6Vpk. That output voltage is clearly impossible, i.e. you can't get 40.96W into 8r0 from an F5 on +-25Vdc.
Now going through the same current peaks for the 4r0 load case.
2Apk into 4r0 is 8Vpk, OK
2.6Apk into 4r0 is 10.4Vpk, OK
3.2Apk into 4r0 is 12.8Vpk, OK
Do you see why we need to check both the power coming from peak currents and the power coming from peak voltages?
Push Pull has another string to it's bow. It transfers over to ClassAB if the demanded output current exceeds the ClassA peak limit.
Let's assume the maximum peak voltage available from a +-25Vdc F5 is ~21Vpk.
The maximum ClassAB Power is 27.56W into 8r0
max ClassAB into 6r0 is 36.75W
max ClassAB power into 4r0 is 55.12W
Power is a combination of current and voltage and a resistance. You must account for all three.
Edited at Andrew's request to correct typing error.
You have to treat Single Ended differently from Push Pull.
You also have to look at the difference that peak current and rms current make to your calculations.
This is an F5 thread, so I'll deal with the F5's Push Pull case.
Bias to 1A allows a Maximum ClassA output current of 2Apk.
Maximum ClassA power is given by Ipk^2 * Rload / 2
For 8r0, this comes to 16W (1AIb)
For 1.3A bias the Max ClassA Power is 27.04W
For 1.6A bias the max ClassA Power is 40.96W
If you feed these same currents into a 4r0 load then the respective maximum Powers are all exactly halved.
But you must also check that the amplifier has not clipped the signal.
2Apk into 8r0 is 16Vpk. That is well within the F5 voltage capability when on +-25Vdc supply rails.
2.6Apk into 8r0 is 20.8Vpk. That is probably right on the nail for maximum unclipped signal.
3.2Apk into 8r0 is 25.6Vpk. That output voltage is clearly impossible, i.e. you can't get 40.96W into 8r0 from an F5 on +-25Vdc.
Now going through the same current peaks for the 4r0 load case.
2Apk into 4r0 is 8Vpk, OK
2.6Apk into 4r0 is 10.4Vpk, OK
3.2Apk into 4r0 is 12.8Vpk, OK
Do you see why we need to check both the power coming from peak currents and the power coming from peak voltages?
Push Pull has another string to it's bow. It transfers over to ClassAB if the demanded output current exceeds the ClassA peak limit.
Let's assume the maximum peak voltage available from a +-25Vdc F5 is ~21Vpk.
The maximum ClassAB Power is 27.56W into 8r0
max ClassAB into 6r0 is 36.75W
max ClassAB power into 4r0 is 55.12W
Power is a combination of current and voltage and a resistance. You must account for all three.
Edited at Andrew's request to correct typing error.