Having decided to embark on my first tube amp project a few questions came up. It seems that there is much information in favor of multifilar wound output transformers, however, I am having a hard time finding enameled or high tech finish copper wire which has an insulation rating above 1000 volts. It may even have to approach 2000 volts. Laying the primary and secondary layers side by side makes the full voltage potential difference only a coat of paint apart.
I could choose to wind one layer of primary and then secondary etc. Then there will be a huge amount of insulation in the stack.
Does anyone have any suggestions? Perhaps teflon fabric as insulation. It is very thin and has good dielectric properties.
The high impedance and voltage needed to drive the 813 tubes is the reason for constructing a multifilar stack. This math is already giving may a major headache. Though I am lucky to have all of the information one could ever need on this subject.
Thanks Tad
I could choose to wind one layer of primary and then secondary etc. Then there will be a huge amount of insulation in the stack.
Does anyone have any suggestions? Perhaps teflon fabric as insulation. It is very thin and has good dielectric properties.
The high impedance and voltage needed to drive the 813 tubes is the reason for constructing a multifilar stack. This math is already giving may a major headache. Though I am lucky to have all of the information one could ever need on this subject.
Thanks Tad
multifilar separate windings minimize leakage inductance at the expense of maximizing parasitc C, I believe that for any high turns ratio other than 1:1 turns ratio this will cut your bandwidth vs balancing the two types of parasitics with interleaving/sector winding techinques
Lenard Audio - Education - Valve Amps shows interleaving (not that I agree with everything on the site)
parallel bifilar is the 1st step in the bunch winding, Litz wire minimizing of skin effect eddy current loss reduction
if you still "need" KV wire-to-wire insulation then search for "triple insulation" magnet wire - 3 layers of differing insulation material (not to confuse with double/tripple/quad "build" - multiple coats of the same enamel/varnish)
alternating single layer windings with added insulating tape uses standard materials
Lenard Audio - Education - Valve Amps shows interleaving (not that I agree with everything on the site)
parallel bifilar is the 1st step in the bunch winding, Litz wire minimizing of skin effect eddy current loss reduction
if you still "need" KV wire-to-wire insulation then search for "triple insulation" magnet wire - 3 layers of differing insulation material (not to confuse with double/tripple/quad "build" - multiple coats of the same enamel/varnish)
alternating single layer windings with added insulating tape uses standard materials
Last edited:
Nowadays it is much easier to obtain teflon insulated winding wire, as used for switchmode transformers to gain safety compliance. The commonest types of insulation are single layer (used for an extra/supplemenary layer of insulation) and triple layer (for full rated compliance of primary to secondary mains). In the case of the OT primary, the single layer insulation type would be most applicable - if it provided an appropriate balance of parasitic capacitance and leakage inductance.
References are Furukawa FWX-E and Tefzel.
Ciao, Tim
References are Furukawa FWX-E and Tefzel.
Ciao, Tim
It seems for my particular application side by side winding of these transformers may not provide the overall improvement I was trying to achieve. I shall begin by winding a rather simple unit and test just what the results are.
Wanting to make this amp ultralinear is going to require some more complex primary stacks than usual. A continuing learning process sometimes with prodigious amounts of magic smoke. The success is worth the effort.
Most go find some Teflon magnet wire -- cheap. It sounds like a good place to start even for conventional winding arrangements.
It would be nice if Champ amps could provide the transformer winding specs. He has done such a good job of capturing my interest with the information he has already provided. He no longer accepts email.
Thanks for the help. Tad
Wanting to make this amp ultralinear is going to require some more complex primary stacks than usual. A continuing learning process sometimes with prodigious amounts of magic smoke. The success is worth the effort.
Most go find some Teflon magnet wire -- cheap. It sounds like a good place to start even for conventional winding arrangements.
It would be nice if Champ amps could provide the transformer winding specs. He has done such a good job of capturing my interest with the information he has already provided. He no longer accepts email.
Thanks for the help. Tad
hey-Hey!!!,
you speak of needing a high impedance for your 813 application and that leaves only a portion of the secondary available to the benefit of a bi-filar wind. Consider winding the two coils separated by a layer if insulation 'paper', same pitch, same number of turns instead of side-by-side in the same layer. You'll get most of the capacitive coupling benefit you seek as well as most of the reduction in leakage L.
cheers,
Douglas
you speak of needing a high impedance for your 813 application and that leaves only a portion of the secondary available to the benefit of a bi-filar wind. Consider winding the two coils separated by a layer if insulation 'paper', same pitch, same number of turns instead of side-by-side in the same layer. You'll get most of the capacitive coupling benefit you seek as well as most of the reduction in leakage L.
cheers,
Douglas
Go ahead, it won't work.
Because the winding ratio is not 1:1 (but let's say 35:1 for a high impedance application) the capacitive coupling is always working against you, no matter when it's done bifilarly or by layers.
"The same number of turns" per layer means rather thin wire for the secondary. Unless you wind a huge stack of primary / secondary layers with parallelled secondaries the DCR of the secondary will be high, bad for damping / output impedance. Winding the huge stack will lead to much capacitance, goodbye treble.
Trick is to find the right balance between capacitive coupling / leakage L to reach the acceptable bandwidth.
The capacitive coupling does *NOT* work against you always. Consider the first turns after the connection to B+ ( at AC ground ) and the turns of secondary starting from DC ground the coupling will work *FOR* you.
As to the same number of turns requiring thin secondary wire, that is again not required; keep the primary at the same pitch as the secondary...you'll have space between the primary wires axially on the bobbin.
cheers,
Douglas
I think the idea may have been for multiple layers, but where each layer is a 1:1 filar mix of primary and secondary - the primary turns in each layer connected in series to the next layer - the secondary turns in each layer connected in parallel with the other layer secondaries. Either the primary, or the secondary turns, would use insulated wire (but not both - probably the primary). The diameter of the insulated primary would be the same as the enamelled secondary.
yes, you have to add the assumption that a tube output xfmr is "impedance matching" and is driven from the high source resistance of the output tube' plate - then increased pri-sec C rolls off the drive V
if you have low driving source impedance at the primary, not "matching impedance" then you can get somewhat better bandwidth with multifilar/high interleave as long as you don't mind the "waste" of driving high current through the pri-sec C shunting the driver
if you have low driving source impedance at the primary, not "matching impedance" then you can get somewhat better bandwidth with multifilar/high interleave as long as you don't mind the "waste" of driving high current through the pri-sec C shunting the driver
Last edited:
I was really looking forward to building this amp with a minimal of tubes. The 813 has such incredible drive. However, with much of the aforementioned information I might be better off driving 4-6 KT88 or the outputs in the Mac MC3500. The extremely high voltages and high impedance required for the 813 tube may well overload my knowledge base on this subject.
There is quite a bit of very good useful transformer data on the web with some rather extensive math. I was hoping to find the winding data I needed without engineering the entire output section. I think I have the power transformer part pretty much in hand.
I will continue to research this output transformer construct and see where it leads.
Tad
There is quite a bit of very good useful transformer data on the web with some rather extensive math. I was hoping to find the winding data I needed without engineering the entire output section. I think I have the power transformer part pretty much in hand.
I will continue to research this output transformer construct and see where it leads.
Tad
pieter, I generally agree with you - the increased coupling capacitance is a distinct problem. Can you indicate whether your experience relates to very close coupling of the primary wires to the secondaries (as per enamelled wire in a bifilar arrangement)? One benefit of the insulated wire is an increased separation (even though the insulation has a higher permeability). Another possible difference between your experience is the winding configuration - for example, the proposed winding may be done with a trifilar - where 2x insulated primary are wound with 1x secondary - which would nominally halve the coupling capacitance compared to the 1:1 filar arrangement.
Ciao, Tim
Ciao, Tim
Hey Tad,
I have a pair of McIntosh MC3500 / MI350 output transformers that are just sitting around taking up space. If you are interested, PM me with an offer...
Daniel
Presume we have to wind the 813 output transformer. For this tube in triode 10k4 related to 8 ohms would be a sensible impedance ratio (winding ratio 36:1).
With your proposed trifilar winding configuration using insulated wire we have to wind a stack of 18 layers, whereby all primary windings are connected in series, and all secondary windings are parallel connected. This means we have 17 nodes where capacitive coupling from primary to secondary ground will compromise the HF bandwidth (one of the 18 nodes, B+, will not harm as it sees secondary ground). These 17 nodes are too many for this pretty high impedance application. A 6C33 might get away with this, not the 813.
A transformer like this will show HF loss under 20 kHz (but because of the distributed coupling it will be without HF resonance
). IMHO the only way to achieve acceptable results is a balanced configuration of primary and secondary sections not based on bifilar or trifilar winding techniques. For high impedance tubes like the 813, but also 845, GM70, 211 and others the number of sections will be less than for tubes like 2A3, 300B with Rp under 1k.
It is worthwhile attempting to clarify the capacitive coupling comparison - in a very simple way - for a better understanding and insight.
A filar winding configuration would have each turn of the primary effectively placed next to the same wire 'length' of secondary (both sides of secondary wire would see a primary wire. Each layer could be wound so that secondaries are positioned directly above each other to minimise intra-layer coupling (simple view only).
A layer winding configuration (Interleaved secondary between two sections of primary)would have each secondary turn next to primary (under and over layers) with an effective length of the secondary wire.
A very simplisitc comparison would see that the interleaved layer configuration would have 1/18 the capacitive coupling (using Pieter's 18 layer example), which would then be about 1/4 the high frequency bandwidth. Of course layer insulation distance etc, etc all contribute to making a comparison not so simple.
The only simple way to improve the filar config (as I see it), would be to use say four or eight insulated primary wires to each secondary wire - on each layer - the separate primary windings would then be connected in series. Then it is back to the old juggling competition.
Ciao, Tim
A filar winding configuration would have each turn of the primary effectively placed next to the same wire 'length' of secondary (both sides of secondary wire would see a primary wire. Each layer could be wound so that secondaries are positioned directly above each other to minimise intra-layer coupling (simple view only).
A layer winding configuration (Interleaved secondary between two sections of primary)would have each secondary turn next to primary (under and over layers) with an effective length of the secondary wire.
A very simplisitc comparison would see that the interleaved layer configuration would have 1/18 the capacitive coupling (using Pieter's 18 layer example), which would then be about 1/4 the high frequency bandwidth. Of course layer insulation distance etc, etc all contribute to making a comparison not so simple.
The only simple way to improve the filar config (as I see it), would be to use say four or eight insulated primary wires to each secondary wire - on each layer - the separate primary windings would then be connected in series. Then it is back to the old juggling competition.
Ciao, Tim
It seems to me that using similar insulation thickness on the secondary wire (relative to interlayer insulation), with multifilared primary wires enclosed around it, can be made equivalent in distributed capacitance to the conventional layered insulation approach.
But the leakage inductance needs to be looked at carefully then. The secondary wire has a short local magnetic leakage path around thru the thick wire insulation. The conventional layered approach has the wires snugged together for each layer so that magnetic leakage pathes have to travel the length of the layer. This makes for long magnetic path length with low leakage inductance. With a long E lamination this can be further lengthened until the leakage path is nearly the same length as thru the lamination, so most flux will stay in the lamination where it is wanted.
The bifilar or multi-filar approach seems only useful to me if the windings have similar AC and DC structure so that the insulation can be kept to a minimum. Only way around would be to use common mode inter-connect chokes on each of the multi-filar windings so that they can be common moded in the layup. This would really eat up the winding space then for the common mode choke windings. And a much higher resistance total too.
But the leakage inductance needs to be looked at carefully then. The secondary wire has a short local magnetic leakage path around thru the thick wire insulation. The conventional layered approach has the wires snugged together for each layer so that magnetic leakage pathes have to travel the length of the layer. This makes for long magnetic path length with low leakage inductance. With a long E lamination this can be further lengthened until the leakage path is nearly the same length as thru the lamination, so most flux will stay in the lamination where it is wanted.
The bifilar or multi-filar approach seems only useful to me if the windings have similar AC and DC structure so that the insulation can be kept to a minimum. Only way around would be to use common mode inter-connect chokes on each of the multi-filar windings so that they can be common moded in the layup. This would really eat up the winding space then for the common mode choke windings. And a much higher resistance total too.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.