One last thing. I have had a look at the picture of the EML20 and it seems to me that nearly half of the (parallel) filament wires is visible outside the plate surrounded by the grid wires. Adding to this the finite size of the plate which will bend E-field lines (boundary effect that will bend the curves well inside the plate even in case of total confinement of grid and filament inside the plate) I just don't believe that there is temperature gradient on the plate to worry about. On top of this they specify that it can be heated AC or DC....
No you don't need them for DC heating. I don't use them. They actually might create more troubles. I think Rod explained that very well in last post.
Get good IT's with large bandwidth instead of wasting time and money. Not for the FR itself but because it will be easier to make it work properly without requiring extra anode current to tame the transformer resonance too close to a typical frequency range of interest. A resonance at 250-300 KHz like some bilifilar IT'sis not a
Get good IT's with large bandwidth instead of wasting time and money. Not for the FR itself but because it will be easier to make it work properly without requiring extra anode current to tame the transformer resonance too close to a typical frequency range of interest. A resonance at 250-300 KHz like some bilifilar IT'sis not a
IME, if you the Monolith then forget the Lundahl's and concentrate in achieving optimal drive (including the source resistance). Not that Lundahl's are bad but Monolith are at least one step ahead. Bifilar windings done right make all the difference in keeping shunt capacitance as low as possible achieving first resonance around 250-300 KHz! Pretty flat load at least up 100KHz and bandwidth of 140-150 KHz or better.... The easier HF load reduces current demand on the driver. That's what I meant....
I think the reason why Lundahl don't go this way is that with full bifilar windings you can't do SE
P and they prefer to have multi-configurable IT's. IME, best Lundahl for your application is the older LL-1635 that can achieve up to 60-70 KHz bandwidth in optimal conditions. Not surprisingly this can't work as SEP....unless one cheats a little bit in special cases....
I think the reason why Lundahl don't go this way is that with full bifilar windings you can't do SE
P and they prefer to have multi-configurable IT's. IME, best Lundahl for your application is the older LL-1635 that can achieve up to 60-70 KHz bandwidth in optimal conditions. Not surprisingly this can't work as SEP....unless one cheats a little bit in special cases....
This completely agrees with my experience. Once you go bifilar (if you are able to deal with its voltage limitations) you will never be able to achieve the same phase/gain balance with another construction. Even after going through every permutation of winding configuration on the Lundahls, never got the performance I wanted. Some configurations were better than others, and I was able to pick the 'best' of the wirings, but still found phasing issues from side to side starting around 12-14kHz.
Yeah, I re-agree Zigzagflux!
Although I have been having in mind to play with the LL-1635/5mA as split load for a power concertina driver to get push-pull drive at the secondary I will certainly end up using it for SE amp.
The PP idea is quite simple: grid resistor, plate supply and 1/2 6BX7 is all I need! In fact with about 220V supply, the 500R DC resistance of each primary winding will cause 14V drop with 28mA. 190V/-14V for the 6BX7 corresponds exactly to 28mA anode current. Bingo! So I just refer the grid to ground through grid resistor and that's it. The absence of DC imbalance will leave a lot of headroom including 5 mA to manage some power tube grid current. Each primary winding will have 32.5H inductance so the max swing I can get down to 20Hz is 81 V rms per side with 28mA (in A1 or AB1 with no grid resistors at the secondaries). Of course this is the max and will start to clip but 50 Vrms + 50 Vrms should be quite possible with very low distortion thanks to local fb. One can drive quite a number of power tubes with that swing. Of course the input stage will have to swing a lot of volts but the input impedance of the 6BX7 in this configuration will be very high (again thanks to fb that basically multiplies the grid resistor). Anyway for class A1 or AB1 using sweep pentodes with CFb the PP version is a better choice having almost 3x inductance. High frequency remains a concern however if it works I might end up doing it with a better bifilar IT.
I have got a good number of NIB 49's that have been waiting in their box for about 60-70 years. 4K source impedance + input capacitance of power tube + stray should match pretty well the manufactures recommended load for flat response achieving -1dB @60KHz. In terms of available swing it should be possible to get clean 60V peak with some 2H cancellation if I decide to go for 2A3-40 for about 7W A1 output (so approx. 25W plate dissipation for long tube life and no stress) or use 10% Schade fb with something like 6GE5 for the same 7W Pout. In the latter case using 10K-90K partition like the original will load the 49 more especially at low frequency but with tailored working conditions for the 6GE5 the necessary swing will be in the region of 23V peak which would require about 3mA peak @20Hz out of 6 mA DC quiescent current. I might likely go for this as I am seeing that properly selected 2A3-40's aren't so cheap....
Although I have been having in mind to play with the LL-1635/5mA as split load for a power concertina driver to get push-pull drive at the secondary I will certainly end up using it for SE amp.
The PP idea is quite simple: grid resistor, plate supply and 1/2 6BX7 is all I need! In fact with about 220V supply, the 500R DC resistance of each primary winding will cause 14V drop with 28mA. 190V/-14V for the 6BX7 corresponds exactly to 28mA anode current. Bingo! So I just refer the grid to ground through grid resistor and that's it. The absence of DC imbalance will leave a lot of headroom including 5 mA to manage some power tube grid current. Each primary winding will have 32.5H inductance so the max swing I can get down to 20Hz is 81 V rms per side with 28mA (in A1 or AB1 with no grid resistors at the secondaries). Of course this is the max and will start to clip but 50 Vrms + 50 Vrms should be quite possible with very low distortion thanks to local fb. One can drive quite a number of power tubes with that swing. Of course the input stage will have to swing a lot of volts but the input impedance of the 6BX7 in this configuration will be very high (again thanks to fb that basically multiplies the grid resistor). Anyway for class A1 or AB1 using sweep pentodes with CFb the PP version is a better choice having almost 3x inductance. High frequency remains a concern however if it works I might end up doing it with a better bifilar IT.
I have got a good number of NIB 49's that have been waiting in their box for about 60-70 years. 4K source impedance + input capacitance of power tube + stray should match pretty well the manufactures recommended load for flat response achieving -1dB @60KHz. In terms of available swing it should be possible to get clean 60V peak with some 2H cancellation if I decide to go for 2A3-40 for about 7W A1 output (so approx. 25W plate dissipation for long tube life and no stress) or use 10% Schade fb with something like 6GE5 for the same 7W Pout. In the latter case using 10K-90K partition like the original will load the 49 more especially at low frequency but with tailored working conditions for the 6GE5 the necessary swing will be in the region of 23V peak which would require about 3mA peak @20Hz out of 6 mA DC quiescent current. I might likely go for this as I am seeing that properly selected 2A3-40's aren't so cheap....
Last edited:
Bifilar gives the best specs but has, as already mentioned, the voltage limitations.
Lundahl IT's are at the other side of the spectrum: a limited number of primary and secondary sections, each consisting of multiple layers.
This gives the flexibility (SE-PP), but is not optimal for a 1:1 SE application.
I just finished a series of IT's for EML20B, but instead of bifilar I wound them layer for layer P - S and so on. Very tight coupling and very low leakage inductance. Almost as much capacitance as bifilar. A bit more work
.Result of this is almost the same bandwidth as bifilar but with high voltage insulation, so less risk of failure.
With 3k3 source impedance 60 kHz -1dB; 170 kHz -3dB; primary inductance 120 H; primary and secondary DCR 339 ohm (nicely some 10% of tube Rp).
No reason at all to go the bifilar route; the specs of these non-bifilars are much better than any SE output transformer so they don't limit the bandwidth of the amplifier.
Bifilar gives the best specs but has, as already mentioned, the voltage limitations.
Lundahl IT's are at the other side of the spectrum: a limited number of primary and secondary sections, each consisting of multiple layers.
This gives the flexibility (SE-PP), but is not optimal for a 1:1 SE application.
I just finished a series of IT's for EML20B, but instead of bifilar I wound them layer for layer P - S and so on. Very tight coupling and very low leakage inductance. Almost as much capacitance as bifilar. A bit more work
Result of this is almost the same bandwidth as bifilar but with high voltage insulation, so less risk of failure.
The bifilar Monolith have 375V prim.-to-sec DC differential SAFE limit. The Sowter 9525 has 300V and less bandwidth but quite cheaper and affordable for what it has to offer. Both tested and not rated with a leap of faith. No risk, or at least is the same as any other transformer. Both good for a lot of applications....
Lundahl's in SE applications are good for those who can use them in optimal way.
At lot of insulation means very likely oversized transformers. Not everyone will like that....
That sounds more like the innkeeper speaking of his wine.
The Monolith IT-03 is non-bifilar, has 750V SAFE limit for primary-to-secondary DC voltage and gets 90 KHz @-1dB with 3.3K source imp. It comes in different flavours both SE and PP application with different or no air-gaps.
However that is not even remotely close to bifilar SE Monolith that reach 100KHz flat in the worst case scenario. With low source impedance and circuit stray cap under control they roll-off at 350-400 KHz, depending on the model. Up to 1MHz for PP types.
The only downside is that one has to know in every detail what is doing.....
The LL-1635 has 60KHz -1dB if used with similar (4K) source impedance...even better for the 20mA version.
Then there are the Tango NC-20, always a reference, and its recent brothers/replacements:
http://www.tube-amps.net/images/Hashimoto_Specs/A-115 .jpg
http://www.tube-amps.net/images/Hashimoto_Specs/A-305.jpg
The dummy secondary load is not necessary by default when actual power tubes are on and low frequency roll-off gets better....
So there is choice.
Last edited:
No; it was to show that there is a very good "midway" between bifilar and Lundahl-style multi-section wound 1:1 interstage transformers as a reply to zigzagflux.
My customer asked for an IT for EML20B, which runs at 32 mA at least (or more in mid or high bias).
His specs: 16k:16k (1:1); 35 mA primary DC current: 160 VRMS at 20 Hz with low LF distortion (hence high primary inductance).
MM IT-03 does not meet these specs: not enough primary DC current capability for their 25 mA model; not enough VRMS at 20 Hz for all IT-03 types; not enough primary inductance for their 60 mA type.
The Tango NC-20 and Hashimoto's, though excellent IT's, don't meet the required specs.
You know perfectly well that these specs are interrelated; my IT meets the customer's demand. With IT-03 specs the IT would have equally good HF specifications.
You cannot compare apples with pears
Last edited:
Again no.
A bifilar or P-S layer to layer wound IT will have exactly the same amount of insulation.
The bifilar wound transformer needs insulation between each bifilar layer, or your transformer will not be bifilar wound anymore after a number of layers
.The concept of bifilar winding in audio transformers is a bit overrated IMO and E.
Actually, in a practical transformer, the coupling is not purely "bifilar", as multiple windings are next to each other. In other words, the primary wire does not "see" only his secondary brother, but also the secondary of the "next" and/or "previous" winding.
Purely bifilar is only the case when the primary winding couples to only one secondary winding, but that is not the case in real life transformers.
You wrote "the best and bla bla" I only replied that's not the case showing examples that match or or do better. Having said this I have only read statements from you and never seen a real measurement of your transformers except some in the net where you really don't get a good review.
My customer asked for an IT for EML20B, which runs at 32 mA at least (or more in mid or high bias).
His specs: 16k:16k (1:1); 35 mA primary DC current: 160 VRMS at 20 Hz with low LF distortion (hence high primary inductance).
MM IT-03 does not meet these specs: not enough primary DC current capability for their 25 mA model; not enough VRMS at 20 Hz for all IT-03 types; not enough primary inductance for their 60 mA type.
The Tango NC-20 and Hashimoto's, though excellent IT's, don't meet the required specs.
You know perfectly well that these specs are interrelated; my IT meets the customer's demand. With IT-03 specs the IT would have equally good HF specifications.
You cannot compare apples with pears
Those are just examples. MM can and will do custom transformers...better than you for the simple reason that they know what they are doing before winding a transformer. Not simply trial and test.
Last edited:
Again no.
A bifilar or P-S layer to layer wound IT will have exactly the same amount of insulation.
The bifilar wound transformer needs insulation between each bifilar layer, or your transformer will not be bifilar wound anymore after a number of layers
The concept of bifilar winding in audio transformers is a bit overrated IMO and E.
Your transformer is oversized generally speaking. You always miss the point that the Monolith is an OFF THE SHELF transformer. All that insulation with those low frequency specs cannot be achieved with 2Kg transformer (including the metal case) like the IT-05, in my book.
The small IT-05 will be capable of 60-65 Vrms @20Hz (i.e. 170-180 V peak-to-peak) with the EML-20 running at 25 mA with rather low distortion. Which is pretty darn good in plenty of application and the 375V pri.-to-sec. SAFE limit will be enough.
Overrated is only your transformer until you don't show real numbers. Once you show that I can say you are right otherwise it just the innkeepr talking of his wine.
Oh real life!!!...Have you got real life measurements of yours for once?
Last edited:
Remarkable then that, at my age of 68, I am still full time busy custom winding transformers
Yes you have not idea of running a business with off-the-shelf products. It IS more difficult!
No, it is not over sized. It is made to meet the customer's demand; it is a little under 4 kg potted, perfectly in line with the weight of an IT-03 considering the differences in specs.
I also don't miss the point that the IT-03 is an (excellent) off the shelf transformer; it just would not meet the customer's demand!
Monolith would be perfectly able to wind to the customer's demand as well
.So comparing my custom made EML20B IT with a stock IT-03 does not make sense.
You finally get this??
Last edited:
I do have an idea as I wind transformers for some 20 years now
In the past I considered "setting up" some standard range, but never reached the point to do so apart from the fact that I never found enough time to set up a standard range (I guess you rather have no idea....).
In practice people ask for different specs, even for transformers for popular projects like 300B SE amps. Primary impedances from 2k5 to 5k and everything in between; secondary impedances from 4 to 16 ohm and everything in between (or multiple secondaries); standard 300B transformers for 60 mA primary or transformers for "super" 300B's with primary current up to 120 mA.... All this varying parameters mean that a "standard" transformer might suit, but quite often does not.
No, it is not over sized. It is made to meet the customer's demand.
I also don't miss the point that the IT-03 is an (excellent) off the shelf transformer; it just would not meet the customer's demand!
Monolith would be perfectly able to wind to the customer's demand as well
So comparing my custom made EML20B IT with a stock IT-03 does not make sense.
You finally get this??
It's you missing the point sorry. It would be oversized for a 300B PP like the subject of this thread, for example.
The discussion was not about SE transformers, I just made an example on how a specific SE transformer could be used for PP drive. But for the rest was and is about high frequency ONLY and the importance of bifilar winding that until today it the only kind of transformer I have found OFF THE SHELF that will have no phases issues. The examples I made were mainly related to high frequency, not specific applications. Instead you only seem to be interested in claiming how good are your SE transformers.... with no real proof!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.