• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Capacitive v. Inductive inter-stage coupling

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am building my first tube amp, and I need to make a decision between capacitive coupling and using an LL1660 transformer for inter-stage coupling. I have heard both positive and not so positive about each. I plan on driving a 300b in A1 operation.

Any comments?
 
Administrator
Joined 2004
Paid Member
There are always tradeoffs... I'm still in the process of trying to answer this question myself. Currently working on a GM70 driven by IT coupled triode connected D3A..

IT drivers generally require tubes with relatively low rp if any sort of decent bass performance is desired. In many cases these same tubes work very well with choke loading and capacitive coupling.

Types like the 6SN7 often chosen for driving 300B, a choice I made 12yrs ago, and still live with today are pretty marginal for IT driver application.. (RC coupled, my driver is SRPP based)

So you need to think about what your goals are and what sort of driver complexity you are looking at.

As a personal note I was very suspicious of transformers at one point, today I have SUTs stepping up my SPU, and plate to line transformers in my 26 line stage. My digital hardware is all transformer coupled as will be the next power amp.. And no regrets..

I'd try the IT approach if you have the design chops for it. D3A, C3G, and 7788 are all types to consider and have great linearity in triode driving IT (or anything else for that matter).
 
That is good advice, i.e., I think I will simplify first. I can start with capacitive coupling, but keep room in my chassis to also include the IT. Then, when the first design is finished, I can modify the mechanicals, so I can listen to one for a while, then the other. Thanks for the input, it gave me some perspective.
 
If you read up on tests that Lynn Olsen did on Nutshell hifi you'll see you get less harmonics with inductors. Therefore there is proof inductive coupling is better.

However inductors have their own physical and cost issues - so you may consider shifting around your PSU lines so you end up not needing any coupling capacitors. Usually there are only two of them to dodge (3 stage amp) so you need two extra rails (+ve and -ve) - or design a PSU ladder than you can hang the circuit off.

Although PSUs have capacitors too, you can also use followers etc, and also the PSU caps can be nicely bypassed with better caps too - PSRR also helps you here.
 
I've gone through many many versions of a 300b SET, and tried out all sorts of coupling. I eliminated plate resistors because they didn't sound so good as iron and reduced the gain. I eliminated active loads because to me they had a glare and hardness despite their clarity and good bass.

That left me with the choice of interstages or plate chokes. In practice you can't really get 1:1 interstages to work too well with plate resistances over about 6k - you need the inductance. Yes, the LL1660 is one choice. But my input tube of choice is the 26 and I prefer that with a plate choke. In fact I use two Hammond 156C in series so I have 300H inductance. That's beyond interstages.

Coupling the driver to the 300B is great for an interstage - just choose a driver with Rp of 5k or under. Lundahl LL1660, or LL1692A is nice, or I also used and liked Hammond 126C, bifilar wound.

Logic told me to direct couple the 26 to the 46 driver I use, but I didn't like the 46 with a bypassed cathode resistor. I much preferred the 46 with battery bias on the grid (24v from two 23A). So that meant a capacitor coupling the stages. I use a Russian FT-2 or FT-3 teflon. Normally I'd try to avoid capacitor coupling but in this instance it's what gave me the best sound. I found that I liked the sound best when the cathode was grounded, or with the 26 just a 10 ohms resistor in filament bias.

So you have a few options to weigh up which are inter-dependent. Some circuits require a cap, some don't. Some tubes like interstages, others like plate chokes. But if it were me I'd avoid all resistors except for the 300b which I still have in self bias since it's safe and reliable and 300bs are expensive!

Andy
 
Andy,

Interesting read as I am now in the process of comparing the sound of RC vs. LC vs. IT mSelf in a pretty straightforward 300B Push-Pull setup (e182cc as driver in LTP with cascoded ccs in the tail and 300B stage with fixed bias.

My impressions so far (current through the tubesalways constant):

- RC: You hear the effect of different resistor brands, cap brands etc. All have their sound. If you are in sound tuning business, lots of choices. I tried Duelund Cast, Cu and Cu-Ag, Jupiter and many more. Duelund giving you a nice midrange, but a bit fuzzy soundstage, Jupiter much more refined, transparent and natural. Mils very sparkling, but mechanical, Caddock warm, but not very dynmaic, Ohmite Gold a good compromise etc etc.

- LL1668AM instead of resistor: More energy, naturalness, but in combination with Duelund now to warm and fuzzy. Combined with Jupiter: Very, very good. The Jupiter is necessary to show the potential of the anode chokes. Compared ro resistor: More natural, warm and dynamic at the sametime.

- Monolith ITA-02 special version with two secondaries to do fixed bias: Still in burn-in...but sofar even more natural, louder and dynamic. Now you can even hear if the coleman raw DC supply for the 300B has Elna Silmic or Epcos lytics inside. Silmic give better tone colour/warmer without taking information away, Epcos, very clear, but a mechanical, stressed.

I ask myself: What to do with my four silk grid chokes ? Put them between the bias supply and the secondaries of the IT ?

I am happy to learn your view on the topic as of today...next step will be to compare SE output stage vs. PP output stage, ideally with the same IT, just with differently configure secandaries, nit to confuse the sound difference of PP vs. SE as a concept with the sound of different driver designs.

All the best
 
Last edited:
IT drivers generally require tubes with relatively low rp if any sort of decent bass performance is desired. In many cases these same tubes work very well with choke loading and capacitive coupling.

If 1:1 ratio is desired, a very high inductance IT can be built. My latest experiment was a non-bifiliar amorphous core IT with a L=150H and bandwidth up to 100kHz at -3dB. Winding of course was a PITA. Price will be an evil factor here. Biliar winding would result in better parameters. In summary, it's possible to use medium to high Rp tubes, this one was intended for VT-25.

Inductance coupling can give very little distortion with good selection of core materials, optimal transformer for the circuit. And it gives the plus to provide a low impedance drive to the power tube, doesn't suffer from RC recovery, but... price.
 
Last edited:
I am voluteering on your transformer...as I still enjoy your fabulous amorphous chokes...

I am going to build a driver based on a EML 20b next...mu of 20, DHT, 3,5k approx....so quiet the same like the 801A when you have a good copy.

Ideally an IT which has 450V DC primary, where the Monolith bifilar stops at 375VDC. and ideally usable for PP-->PP as well as PP-->SE....which the bifilar cant do as well...

If this is feasible, drop me an email... :)
 
i will choose Direct Coupling any days (but you have to deal with high voltage)

This has a lot of virtues, but is certainly not a universal solution. In PP you still need a phase-splitter and applying fixed bias through dc coupling appears risky.

It is not like the transformer option is straightforward or easy either. Once you outgrow Lundahl, the choice in available iron is very, very limited.
 
I've gone through many many versions of a 300b SET, and tried out all sorts of coupling. I eliminated plate resistors because they didn't sound so good as iron and reduced the gain. I eliminated active loads because to me they had a glare and hardness despite their clarity and good bass.

That left me with the choice of interstages or plate chokes. In practice you can't really get 1:1 interstages to work too well with plate resistances over about 6k - you need the inductance. Yes, the LL1660 is one choice. But my input tube of choice is the 26 and I prefer that with a plate choke. In fact I use two Hammond 156C in series so I have 300H inductance. That's beyond interstages.

Coupling the driver to the 300B is great for an interstage - just choose a driver with Rp of 5k or under. Lundahl LL1660, or LL1692A is nice, or I also used and liked Hammond 126C, bifilar wound.

Logic told me to direct couple the 26 to the 46 driver I use, but I didn't like the 46 with a bypassed cathode resistor. I much preferred the 46 with battery bias on the grid (24v from two 23A). So that meant a capacitor coupling the stages. I use a Russian FT-2 or FT-3 teflon. Normally I'd try to avoid capacitor coupling but in this instance it's what gave me the best sound. I found that I liked the sound best when the cathode was grounded, or with the 26 just a 10 ohms resistor in filament bias.

So you have a few options to weigh up which are inter-dependent. Some circuits require a cap, some don't. Some tubes like interstages, others like plate chokes. But if it were me I'd avoid all resistors except for the 300b which I still have in self bias since it's safe and reliable and 300bs are expensive!

Andy
Can't agree more. Andy, I went through many of the versions you described. I've build a 26-46-50 and I liked the LL 2753 as IT for the 46. I found that the 26 sounds best with Ale's plate gyrator and filament bias. On the other hand, I like very much the sound of a C3g loaded on 1660 to drive the 300B. There is not much headroom, but if one is happy with low power levels than it should be a great sounding amp and less complex.

Cheers,
Radu
 
It is not like the transformer option is straightforward or easy either. Once you outgrow Lundahl, the choice in available iron is very, very limited.

Unless you are going for custom IT transformers best practice is to design around the transformer. Given the great variety of drivers one can use this is no problem. The problem is mystical suggestion people have about this and that tube....:D
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.