Bipolar Cap Mult DC Heaters

[atavacron]

I dont like grounding one side of a DC heater supply in tube gear ,
I prefer to tie each side of the heater to ground or its reference voltage via resistor center tap and have its own discrete +/-connection back to the psu .
50 ohms each leg and its worth using the meter to select a matched pair .
I place that at the PSU output end with a very good 0V connection direct to chassis ,

I will have to think on this, thank you for a detailed robust option.

 

[abbey road d enfer]

I prefer to tie each side of the heater to ground or its reference voltage via resistor center tap and have its own discrete +/-connection back to the psu .
50 ohms each leg and its worth using the meter to select a matched pair .

It makes sense with AC heaters, not so much with DC heaters.
There are many examples of grounded DC heaters that work perfectly.
With AC heaters, best noise rejection is not necessarily when resistors are perfectly matched. That's why the best designs use a "hum-balance pot".

I place that at the PSU output end with a very good 0V connection direct to chassis ,

This is to dump electrostatic leakage to ground. Doesn't need a very sturdy ground.

 

[ruffrecords]

I suppose I have to qualify what I said by adding it was in a power amp I saw these bad effects ,

maybe less of an issue in preamps ,
still though for the sake of a slightly larger Meanwell or transformer ,why bother with series parallel heaters ,

For a preamp with a small number of tubes the only issue is how easy is it to obtain a decent 6.3V dc supply. If SMPS went that low I would recommend that route but they are not generally available or tweakable to that voltage. It is when you have many tubes that running them in pairs makes sense because it halves the heater current. In my small 6 channel mixers I run all the supplies and buses along a backplane PCB into which the channel modules are plugged. I use a fat track on both sides of the PCB to carry the heaters. In a six channel mixer the heater current at 12V is nearly 3 amps. At 6V it would be twice that. My Mark 3 design has 12 modules so even at 12V the total heater current exceeds 6 amps. SInce I use an external power supply you also have to consider the voltage drop from PSU to mixer. 12V heaters makes sense in this situation.

If I can find time this evening I will make a small 2 tube heater test rig and run random pairs of 6922 tube heaters in series from my bench supply and measure the actual imbalance. I will let you know the results.

Cheers

Ian

 

[abbey road d enfer]

It's important to make a distinction between powering heaters of separate tubes in series and powering a dual-triode such as a ECC8x/12A#7. The latter are expressly designed to allow both parallel and series connection. For separate tubes, there is no guarantee that the resistance/heating current will be identical, even if they are of the same brand.

 

[merlin]

I thought the dropout for a depletion mosfet used as a source follower was just RdsOn * current. No? That part is like $8 in 10 qty btw, you really pay for the low resistance.

Ah sorry, I didn't realise it was depletion, I thought it was an ordinary enhancement device.

 

[SoundBlaster]

Hi All. This is a great use for little DC-DC switcher modules. A number of these can be had that operate in the MHz range. I've used used some LM2596 modules from Amazon (65kHz) Two for $7.00! Up To 35V input, Adjustable output. 5A!. Start with a bulk DC supply that you can put a big cap on. All these source follower circuits that have been suggested or resistors to burn off a few volts do is make things hotter than they need to be and less efficient.

 

[Tubetec]

Almost all the great tube mics have one side of the heater connected to ground ,
I prefer not to use the cable screen as the heater return wire ,
Typically a tube mic cable has a pair of higher gauge wires specifically for the heaters .

Some tubes have a spiral heater , EF86 for instance ,
From the point of view of externally induced noise the +/- and ground heater arrangement can better reject outside sources .
Were in a very different world today in terms of electronic smog

Of course Im familiar with the 100ohm WW pot used as Humdinger used in some vintage amps
when you have multiple tubes each contributing a different level of hum and perhaps some phase cancelation across different stages minimum hum isnt usually at the 50% tapping point .
With a single tube preamp I found the resistor center tap to work well .

 

[ruffrecords]

I did some tests on series wired heaters of pairs of EH6922 tubes this evening, I tested 10 random pairings of 10 tubes. I connected them to my bench supply which was set to 12.1V. I measured the actual voltage across each heater and the current flowing (in all cases the current was 0.35 amps). Here are the test results:

Htr1 Htr 2
6.11 6.05
5.92 6.16
5,99 6.11
6.03 6.06
6.02 6.06
5.99 6.10
6.00 6.09
5.92 6.18
5.9 6.2

As you can see, they are all withing 0.3V of each other and in most cases much closer. For comparison I found a pair of NOS ECC88 tubes and tried them. The heater current was 0.3 amps and the voltages were 5.89V and 6.22V so only slightly worse than the worst 6922 pairing. Finally I tried a pair of NOS Russian 6HN23 which are a supposed ECC88 substitute, This pair also drew 0.3 amps current but the voltages were 6.39V and 5,72V - nearly 0.6V difference.

I conclude that the heater properties of the EH6922 seem to be well controlled in production and it is is quite safe to use pairs of EH6922 with their heaters wired in series.

Cheers

Ian

 

[atavacron]

Thanks Ian! To review:

EH ≈ 350mA
JJ ≈ 320mA

JJs cost less and have lower mu generally, and you have to buy more of them to weed out the bad ones. I’ve asked about bulk discounts on EH, it doesn’t sound like that’s currently happening unfortunately.

I’ll sim at 4 x 350mA @ ≈12.6V and 8 x 350mA @ ≈ 6.3V with the lowest drop DO-201 schottky to be found on Mouser (Panjit SBM1045VSS), just to see what’s possible.

 

[atavacron]

Well this is rather encouraging. (scroll down)

I'm using someone's SB560 model, which per its datasheet would be about -1.06V in a FWB (at 2A), vs my preferred SBM1045VSS at -0.63V in a FWB (at 2A), so the 16.25V we're seeing here would actually be 16.68V.

At 2A, I've got a pass element that drops about 0.6V, and a current limiter that drops about 1.24V, so that's 16.68 – 0.6 – 1.24 = 14.84V coming off of my remote PSU. That's more than enough to allow for a little drop through a five foot 16awg cable (it's a 4-pin XLR), and enough to allow for a regulator on each preamp card to dial down to 12V for the heaters and relays themselves. I'd have to see if I could get away with an LM317 or if I'd have to go to a lower dropout model, but either way it wouldn't be handling any more than 500mA per card.

Also, wow, the difference between a single 1.59Hz RC and an RCRC is quite significant - 0.008V vs 0.0001V ripple. When I ran this with just the single RC and changed my 1uF to 82uF (roughly what the cap mult might accomplish?), it only halved the ripple. Also the difference at the reservoir cap between 2x 8200u and 3x 8200u is about 260mV, which is less worth it than going from 1x to 2x 8200u — that’s about a 750mV advantage.

This all sound like a good starting point?

 

[abbey road d enfer]

At 2A, I've got a pass element that drops about 0.6V, and a current limiter that drops about 1.24V, so that's 16.68 – 0.6 – 1.24 = 14.84V coming off of my remote PSU.

And what about the voltage drop in the secondary winding?
And 2A DC means about 4-5A peak. It's the peak current that governs the voltage drop.

 

[thor.zmt]

Dude calm down, it's two tubes of the same type in series. Millions have been run that way. It's fine.

Yes, millions in Radio's and TV set's that used tubes SPECIFICALLY DESIGNED for series heater connection.

simplicity. same reason i’d rather not do the step-down-for-heaters-and-back-up-for-ht thing.

Simplicity needs to be balanced with electrical safety requirements, availability, cost. Let me propose this:

100VA 9V+9V

but if i can get the single 12.6V supply dialed, i’m gonna take what i learned (namely those schottkys and mosfet from yesterday) and see if i can squeeze out enough volts for all parallel heaters.

I checked, no:



There is totally no headroom for regulation and low line voltage.

also, i’m having a heck of a time finding toroidal options in the US for HT around 170V-180V (pre-rectifier)…even without heater windings.

IF you use this:

115V + 115V : 9V/6.66A + 9V/6.66A (120VA)
12V/2.5A + 12V/2.5A : 115V+115V (60VA)

You get 172.5V HT at all nominal load.

that 175 i’m using doesn’t exist in one size up or down. there’s toroid.com, but they only have two standard models. toroidy.pl is clearly a resource, but with the euro-to-dollar mismatch, i don’t even wanna know the shipping. i’m hesitant to ask for a custom until i’m really ready to order 10+. does anybody stateside come to mind?

If you are looking at small series production, take a BIG step back.

24V Output switchers for LED are plentiful, cheap and small.

24V to 5.7...6.3V DC-DC converters are TRIVIAL to find, design etc. Modern ones run around 1MHz and can be synchronised.

For example the RTQ2965GSP-QA will tolerate 60V in and has 5A Out, can go up to 2.5MHz switching in an SO-8 package with thermal pad (relatively DIY/small series friendly if you put a big hole in the PCB to each the thermal pad with a solder iron tip) and system efficiency (with diode and inductor and capacitor losses) of 85% going 24V -> 5V. Output is adjustable. Only non-electrolytic capacitors would be needed for such a design (reduced aging issues).

Now HT is a bit harder to do. BUT, you can take a low voltage step-up (boost) switcher and use a suitable high voltage switcher Mosfet as cascode on the switching node to allow high voltage.

As these Mosfet's are usually targeted at 100's of kHz switching speeds, a MC34063A as the main switcher part makes sense. To get 240V/0.15A DC from 24V, here what is needed:



MURATA POWER SOLUTIONS 1468420 is a suitable off the shelf inductor. You can probably get away with sensible size MKP capacitors for the HT and additional LC filtering/

So you now have appx. 6V/5A + 240V/150mA from a really compact PCB plus a ~100W/24V Meanwell switcher, with low noise, no issues with 50/60Hz magnetic fields and so on. Total cost is likely a fraction of just the transformer. If this is not enough weight, build the PCB into a really thick steel box.

Thor

 

[thor.zmt]

As you can see, they are all withing 0.3V of each other and in most cases much closer. For comparison I found a pair of NOS ECC88 tubes and tried them. The heater current was 0.3 amps and the voltages were 5.89V and 6.22V so only slightly worse than the worst 6922 pairing.

Finally I tried a pair of NOS Russian 6HN23 which are a supposed ECC88 substitute, This pair also drew 0.3 amps current but the voltages were 6.39V and 5,72V - nearly 0.6V difference.

Now imagine a customer plugs in the higher voltage NOS 6N23 with 0.3A in together with a 0.35A current production one that has low voltage.

Or worse, a rare vintage NOS Siemens 1950's CCa (499 USD a pop).

5.9V / 0.35A for one tube
6.4V / 0.30A for the other

With no change in heater resistance we can expect 0.315A with one tube at 6.7V and the second at 5.3V. But in reality the low tube will be even lower and the high tube even higher.

So I'm sorry, I would not recommend series heaters unless you can be sure on every tube change well matched units in terms of heater voltage go in.

Even IF I am the only one ever the using the Unit, I'd probably only do it with a HUGE warning sticker to remind in a few years when it's time to change tubes. And who knows if by then I can still find well matched current production?

For the benefit of skipping a trivial step-down switcher including with off the shelf from Ali Baba and his 40 UNRWA workers it does not seem a reasonable tradeoff.

Thor

 

[thor.zmt]

This all sound like a good starting point?

Probably not.

You need to allow at the least for a range of 114V - 126A AC. On the safe side I'd permit 110V - 5%, or 105V. And that does not cover Japan.

So at 104V AC all your voltages are only 86% of nominal and you still should be at least on the edge of regulation. On top, thermally you need to design for 126V in.

Thor

 

[ruffrecords]

Now imagine a customer plugs in the higher voltage NOS 6N23 with 0.3A in together with a 0.35A current production one that has low voltage

Tube rolling is, and always has been at the risk of the customer.

The argument does not stand up anyway. The design could easily call for a pair of PCC88s which ARE designed to be run in series. But then there is still nothing to prevent the user replacing one or more of them with a 6922 or a 6N23P

And we don't even know there is a customer. As far as we know this is a one off DIY project.

Cheers

Ian

 

[atavacron]

It’ll be me and one other person doing assembly, 12 units once I get through proto. if they sell, we make another 12. it’s a 1U 4-channel mic/line stage with 2-pole HPF and anode follower trim using the SRPP. the PSU is a floor box.

i updated the supply, will post once sim’d.

it’ll be extremely clear that customers should not tube roll. but they can follow instructions to calibrate a replacement tube, or just remove the offending channel card and send it in.

 

[thor.zmt]

it’ll be extremely clear that customers should not tube roll. but they can follow instructions to calibrate a replacement tube, or just remove the offending channel card and send it in.

Can I be a major PITA?

Ignoring the power source (be it your 100VA transformer or a 12....24V Mean Well Switcher), is there really mileage avoiding a Stepdown switcher (off the shelf or custom) for 2 bux on the heaters and wiring heaters in parallel, so any "compatible" tube can be plugged into any socket (the way they were designed)?

And to be brutally honest, seeing where modern DC-DC converters are, is there a good reason not to use an external COTS 20V/5A USB-C PSU that deals wit agency approval etc. (if needed solder wires to the Pin's and put into into a steel brick if you must pretend)?

And add a stepdown DC-DC for heaters (with a second winding for an isolated heater winding for upper sections of SRPP, Mu-Follower, White Follower et al) and a step-up with charge pump for HT?

Cost, weight, reliability etc. will be on another level.

Thor

 

[atavacron]

@thor.zmt you’re on another level (that’s a compliment). I follow your reasoning, but there can only be two PCBs in the system — the PSU and the channel card — and I am not about to learn how to design a SMPS. Just designing a linear supply with (counts on fingers) seven voltages is enough of a challenge. If the thing gets reviewed well or develops a cult following and I need to make a bunch of them, we can surely revisit the fundamental approach to power.

Is there some issue with parallel secondaries on the same trafo that I don’t know about? I’m doing that for both HT and heaters now that I’ve established that I can get adequate headroom from a full wave bridge. Even 6.3V works down at the end of the chain, though I’ll probably go for 6V. It’s 2.8A plus relays, which I’m calling 3.3A, but will probably be lower

 

[moamps]

...... Not exact the end of the world for hobby electronics...

We discuss here how a device should be designed in the correct way. And the production of 12 devices, with the possibility of more, as it was said in post 56, certainly does not mean that it remains within the scope of the hobby.
Although I myself am not critical about connecting heating from electronic tubes in series or in parallel, it is obvious that the manufacturer has clearly stated how they should be connected and I do not see why it should be done differently in a well-designed device.
An example of how this goes badly is the Tubetech MP1 which has three electronic tube heaters connected in series which are not intended for this, and which also has a catastrophically poorly designed power supply. And some of Paravichini's designs are similar.
And that's why we engineers shouldn't encourage such ideas that are technically not correct. Although they actually work.
P.S.
Here for further discussion, if it needs to be discussed at all, I'm really curious if any of you proponents of series connection of electronic tube heaters have investigated the dynamic change of current and voltage at the moment of switching on (cold start).

 

[thor.zmt]

@thor.zmt you’re on another level (that’s a compliment). I follow your reasoning, but there can only be two PCBs in the system — the PSU and the channel card — and I am not about to learn how to design a SMPS.

For the minimum, do the heater supply with a switcher Stepdown using windings in series.

https://www.ti.com/product/LM2679

They call these "simple switcher" for a reason.

This gives you an adjustable, switching regulator that is as easy to apply as LM317. All parts etc are covered in detail in the datasheet.

Just designing a linear supply with (counts on fingers) seven voltages is enough of a challenge.

Shouldn't be, I mean it's just s power supply.

Is there some issue with parallel secondaries on the same trafo that I don’t know about?

Get the phasing right. But seriously, no need.

I’m doing that for both HT and heaters

The secondary windings are for full wava rectification, use them like that for HT.

Even 6.3V works down at the end of the chain, though I’ll probably go for 6V. It’s 2.8A plus relays, which I’m calling 3.3A, but will probably be lower

With heater windings in series, you get 17V at 2A nominal. That's 34 VA. With 80% efficiency (which is poor) fir the switching Stepdown you get 6V/4.5A.

Time to join the 21st century.

Thor