For real performance OTs:
Long E laminations and split bobbin (twice the length of a standard scrapless E, so two bobbins fit on it lengthwise)
C core with split bobbin
Toroid with progressive wind or micro sectioning
Long E laminations and split bobbin (twice the length of a standard scrapless E, so two bobbins fit on it lengthwise)
C core with split bobbin
Toroid with progressive wind or micro sectioning
Attachments
Last edited:
smoking-amp;2471164 C cores (grain oriiented) would certainly be an improvement over standard scrapless E-I said:
Yes, for SE 6S4S. But I still don't understand why do you need split bobbins: to trade-off leakage inductance for better symmetry?
SPlitting the bobbin is not alway benificial....
Depends on what your doing....
If you have triodes in PP vs having pentodes in PP you have different requirements for the tradeoffs with the parasitics of winding capacitance and leakage.... this also depends on the plate load value....ect..ect... You also need to consider not only Pri-Sec coupling but you also need to consider the Primary half to Primary half coupling....
Chris
Depends on what your doing....
If you have triodes in PP vs having pentodes in PP you have different requirements for the tradeoffs with the parasitics of winding capacitance and leakage.... this also depends on the plate load value....ect..ect... You also need to consider not only Pri-Sec coupling but you also need to consider the Primary half to Primary half coupling....
Chris
"I hope you don't mean to split primary and secondary windings (primary one leg, secondary the other leg)? "
No, definitely not. Each bobbin gets interleave wound with secondary and both side primaries just like a standard OT bobbin. Just that half the total interleaves get put on each bobbin. Obviously one will end up with more total interleaves than a single bobbin setup.
"I still don't understand why do you need split bobbins: to trade-off leakage inductance for better symmetry? "
There is no tradeoff if each bobbin is interleave wound just like a single bobbin OT. There is no separation of windings being used here. In fact, the leakage L will reduce due to the effective doubling of interleaves and the geometric factor of long thin windings which have lower leakage L. Having symmetric windups leads to balance in the OT with regard to leakage L and winding resistance, which will reduce HF distortion over a standard single bobbin setup.
"If you have triodes in PP vs having pentodes in PP you have different requirements for the tradeoffs with the parasitics of winding capacitance and leakage.... this also depends on the plate load value....ect..ect... You also need to consider not only Pri-Sec coupling but you also need to consider the Primary half to Primary half coupling...."
Yes, agree. The split bobbin almost forces more total interleaves in order to get both primary sides present in each bobbin (primary half to pri half coupling). This can be offset by adjusting the layer insulation thickness a bit greater to reduce the capacitance and re-balance the resonance impedance for critical damping by the effective load. Similar issue to putting more interleaves on a single bobbin.
Or one could get more fancy and use progressive winds or Pi windings to lower the capacitance some (but not the winding to winding C that way) There are also some tricks using reverse winding to get winding AC voltage progressions aligned between windings, generally used for the two primary halves. And of course the usual alignment of AC grounded ends of primary windings with the secondary or core capacitance. And not to forget, the Mac bifilar CFB technique which effectively gets rid of half the turns. Circlotron configuration is the ultimate at that.
No, definitely not. Each bobbin gets interleave wound with secondary and both side primaries just like a standard OT bobbin. Just that half the total interleaves get put on each bobbin. Obviously one will end up with more total interleaves than a single bobbin setup.
"I still don't understand why do you need split bobbins: to trade-off leakage inductance for better symmetry? "
There is no tradeoff if each bobbin is interleave wound just like a single bobbin OT. There is no separation of windings being used here. In fact, the leakage L will reduce due to the effective doubling of interleaves and the geometric factor of long thin windings which have lower leakage L. Having symmetric windups leads to balance in the OT with regard to leakage L and winding resistance, which will reduce HF distortion over a standard single bobbin setup.
"If you have triodes in PP vs having pentodes in PP you have different requirements for the tradeoffs with the parasitics of winding capacitance and leakage.... this also depends on the plate load value....ect..ect... You also need to consider not only Pri-Sec coupling but you also need to consider the Primary half to Primary half coupling...."
Yes, agree. The split bobbin almost forces more total interleaves in order to get both primary sides present in each bobbin (primary half to pri half coupling). This can be offset by adjusting the layer insulation thickness a bit greater to reduce the capacitance and re-balance the resonance impedance for critical damping by the effective load. Similar issue to putting more interleaves on a single bobbin.
Or one could get more fancy and use progressive winds or Pi windings to lower the capacitance some (but not the winding to winding C that way) There are also some tricks using reverse winding to get winding AC voltage progressions aligned between windings, generally used for the two primary halves. And of course the usual alignment of AC grounded ends of primary windings with the secondary or core capacitance. And not to forget, the Mac bifilar CFB technique which effectively gets rid of half the turns. Circlotron configuration is the ultimate at that.
Last edited:
+10
The MC3500 was easily the best sounding tube amp of that era. Back in the 70's we did a lot of comparisons and the Mc3500's easily out stepped the others ( marantz, scott, HK, Crown, AR, quad, Altec, ) on the bench only the best of SS stuff tested better , try passing a 20 or 20K square wave thru most tube amps, MC3500 No Prob, absolutely no tilt .
What about 6LQ6 ?
+10 .. a rare moment of agreement.. 😛
Please post test bench results of these tube amps. Interesting to see them better a MC 3500..
Not sure about other Tube Macs, but the re is nothing "bloomy" about an MC3500, fast, dynamic and powerful comes to mind ..
An MC3500 is 350 watts in to 16,8,4,2,1 .. I have benched tested one to as low as 2 ohm. Your Pyramids are 16,8,4, and only 80 watts , interesting to see one bench tested for comparision.
One would think for a PA setup 200 watts min would be necessary, why only 80 watts ? ....
"What about 6LQ6 ?"
Yes, going to a high current tube is a big big plus as far as the OT realization is concerned. I always use sweep tubes. 6HJ5 seems to be the golden price performance performer at the moment, maybe 6GY5 and relatives (there are many, $1 a tube even) soon too.
On the Mac driver "headroom" issue that has been raised, some comment might be useful here. The Mac driver does indeed need more voltage output swing capability to drive the output stage grids in CFB, but not in the usual sense of the driver tube having to provide it. With the bootstrapped load (or using a tracked CCS bootstrap), the driver tube does not have to transition any more of its own grid swing than a normal driver stage (no additional current output required), so distortion in the driver tube is not increased over a standard stage. Just the tube ratings have to allow for the plate voltage excursion. There are plenty of tubes that can handle that. If you want extreme voltage excursion say for a screen grid output drive, then maybe something special like a 6GF5 driver.
Yes, going to a high current tube is a big big plus as far as the OT realization is concerned. I always use sweep tubes. 6HJ5 seems to be the golden price performance performer at the moment, maybe 6GY5 and relatives (there are many, $1 a tube even) soon too.
On the Mac driver "headroom" issue that has been raised, some comment might be useful here. The Mac driver does indeed need more voltage output swing capability to drive the output stage grids in CFB, but not in the usual sense of the driver tube having to provide it. With the bootstrapped load (or using a tracked CCS bootstrap), the driver tube does not have to transition any more of its own grid swing than a normal driver stage (no additional current output required), so distortion in the driver tube is not increased over a standard stage. Just the tube ratings have to allow for the plate voltage excursion. There are plenty of tubes that can handle that. If you want extreme voltage excursion say for a screen grid output drive, then maybe something special like a 6GF5 driver.
Last edited:
The context of my comment appears to have been lost - I was referring to the difficulty in getting adequate swing using a typical tube that is often used as a driver tube, that having a maximum B+ rating in the 300 - 600vdc range, and having to drive a tube with >1kv on the plate that is running as a follower... so in that case positive feedback in the form of bootstrapping the driver tube was used by McIntosh.
On another topic - I'm not sure that the use of a "C" core in and of itself will produce a superior transformer compared to the same winding on an equivalent EI core?
_-_-bear
Well, I don't know a lot about transformers but I do know that magnetic fields don't like corners.
Core loss and field leakage arising from coarser and ill-fitting laminations - oh, and the effective increase in magnetic path length due to those damn corners!
The C core and toroid have the preferred grain orientation in the right direction with the field thru them, so they maintain high permeability up thru high flux levels. The E-I core has the wrong grain orientation at the back of the E. The long E does somewhat better by having more % of the path in the right direction. Then there is the joint between the E and I lams. This performs OK up till about 50% of the max flux, then the remaining flux has to jump the butt gaps. So the permeability curve peaks earlier (and with a lower peak) than the C core or toroid which have broad flatter Mu curves with higher effective Mu.
Result is that less core is needed for the C or toroid for the same power (since the Mu is still useable up to a higher flux level), the inductance will be higher for the low freq. perf. and the shorter wire length resulting (from the now reduced core cross section) will improve the high freq. end some too, due to less dist. capacitance.
Although the C core does have all butt gaps to cross at two points, it is precision ground and strapped under tension, so it comes closer to continuity for the field. One could precision grind an E-I too, but I've never seen it done. Some mass produced E-I xfmrs have the I's all on one side and are hydraulically pressed and welded together, but I suspect that just gets the performance back on par with interleved lams.
There also were some special modified E lams made by an outfit in NJ I think at one time at least. These had twice the usual width to the back of the E lam, and could be used in an I-less mode, where the E's were pushed fully together. This eliminated the butt joints and still provided the same amount of cross material at the back, but still incorrectly oriented.
For the DIYer, there is the option of using all I construction (four sided box with two bobbins, each interleaved and cross wound, ie split bobbin design). This gets the grain orientation all in the correct direction. Just not economical for production xfmrs.
More work to stack the lams too.
Result is that less core is needed for the C or toroid for the same power (since the Mu is still useable up to a higher flux level), the inductance will be higher for the low freq. perf. and the shorter wire length resulting (from the now reduced core cross section) will improve the high freq. end some too, due to less dist. capacitance.
Although the C core does have all butt gaps to cross at two points, it is precision ground and strapped under tension, so it comes closer to continuity for the field. One could precision grind an E-I too, but I've never seen it done. Some mass produced E-I xfmrs have the I's all on one side and are hydraulically pressed and welded together, but I suspect that just gets the performance back on par with interleved lams.
There also were some special modified E lams made by an outfit in NJ I think at one time at least. These had twice the usual width to the back of the E lam, and could be used in an I-less mode, where the E's were pushed fully together. This eliminated the butt joints and still provided the same amount of cross material at the back, but still incorrectly oriented.
For the DIYer, there is the option of using all I construction (four sided box with two bobbins, each interleaved and cross wound, ie split bobbin design). This gets the grain orientation all in the correct direction. Just not economical for production xfmrs.
More work to stack the lams too.
Last edited:
As soon as my 200W monoblock prototype will be on the bench I will carefully test it for you.
One would think that high efficiency line arrays are the way to go for high-end PA, and 80pc is adequate for a typical concert hall with 800 seats.
Wavebourn,
Are you going to give us a look inside this 200 watt prototype? I would love to build one of your amps if I could get hold of a schematic.
Has anyone actually seen the MC3500 output transformers with all of the potting removed. I would be interested to know if they are random wound or neatly stacked windings. I also would like to know what configuration they used. There are many pictures around of Mac solid state trannys and some of the smaller tube models but not the 3500.
Many people also claim that this amp will deliver close to 500 watts. That seems quite a stretch for 8 6lq6 bottles.
Having owned several Mcintosh amps and tuners I can say the sound is quite good but not without some color. The one big item is the build quality. They are indeed designed to withstand decades of use.
Tad
Are you going to give us a look inside this 200 watt prototype? I would love to build one of your amps if I could get hold of a schematic.
Has anyone actually seen the MC3500 output transformers with all of the potting removed. I would be interested to know if they are random wound or neatly stacked windings. I also would like to know what configuration they used. There are many pictures around of Mac solid state trannys and some of the smaller tube models but not the 3500.
Many people also claim that this amp will deliver close to 500 watts. That seems quite a stretch for 8 6lq6 bottles.
Having owned several Mcintosh amps and tuners I can say the sound is quite good but not without some color. The one big item is the build quality. They are indeed designed to withstand decades of use.
Tad
Sure, I keep no secrets. Those who want to steal design would not recognize it's value; those who are able to recognize it's value can design something own. Those who can learn and benefit would do that, and I would be glad to help.
Absolutely true. Very robust and shiny mechanical design, made with love.
smoking-amp. yes, I agree. 😀
Wavebourn, I would be very happy if ur amps performed as well as or better than the McIntosh's do. No problem.
Thanks for noticing that I selected Mosfets in my amp to avoid degeneration resistors. 😀
Not everyone does... (fwiw, I did not need to have a batch of 1,000 to get very very good matches. Phew!)
Fwiw, I know Cerrem online for a long time now, and he has a very very good handle on the specifics of the inner workings of the McIntosh transformer design - far far more than I do.
I am happy to admit to being a non-expert on transformer design. And what I did learn is hyper rusty since I have not done a blessed thing with it for more than a decade now... crud!
_-_-bear
Wavebourn, I would be very happy if ur amps performed as well as or better than the McIntosh's do. No problem.
Thanks for noticing that I selected Mosfets in my amp to avoid degeneration resistors. 😀
Not everyone does... (fwiw, I did not need to have a batch of 1,000 to get very very good matches. Phew!)
Fwiw, I know Cerrem online for a long time now, and he has a very very good handle on the specifics of the inner workings of the McIntosh transformer design - far far more than I do.
I am happy to admit to being a non-expert on transformer design. And what I did learn is hyper rusty since I have not done a blessed thing with it for more than a decade now... crud!
_-_-bear
Let's summarize:
As I said, there are 2 major ways to skin the cat: either parallel feedback like I did, or feedback in series like Mac did. Bot are valid. His way required higher linear voltage drive, mine required higher linear current drive. I like mine better: no need for complex transformers, simple Edcors perform well.
Fine, fine. But since we don't have ur schematics to compare nor your results, we'll just have to take ur word for it. But it is important to note that things that would appear to produce the same "results" or equivalent results, or better results, do not always result in a better sound to the ears of some listeners... but again, I am always happy to find another way to do things.
Also, you can as Cerrem for his opinion as to the complexity of winding the basic McIntosh output transformer. Having not wound any, I can't say that it is more or less difficult. Complexity, it actually seems less than a standard audio transformer. I could be wrong.
_-_-bear
Also, you can as Cerrem for his opinion as to the complexity of winding the basic McIntosh output transformer. Having not wound any, I can't say that it is more or less difficult. Complexity, it actually seems less than a standard audio transformer. I could be wrong.
_-_-bear
- Status
- Not open for further replies.