Fairchild 660 from scratch

[fazeka]

There's nothing really gained by using tubes in the B+ regulator and the same regulator circuit could be done using mosfets. It would save money and heat, or would it be blasphemy?

Not blasphemy, thanks for your design! Just to confirm, looks like J4 = control voltage, J7 = signal amp and J8 = 6V6 grids?

Also, it seems the STW8NK80Z is not readily available from the major suppliers, at least in the US. Some saying mid 2023 availability!

 

[fazeka]

Thanks DaveP for checking in! I really appreciate it!

Re: the Hammond 1645. I would hope that the schematic on their site is up-to-date (I've posted it a few posts back) but you're right, sometimes what happens in production doesn't manifest itself into the documentation until (much?) later...

That said, do you know how it is wired differently now?

Thanks,
Chris

The Hammond 1645 that I used was old stock, I believe that the secondary is wired differently now, check out before purchasing!

 

[fazeka]

Thinking about Heikki's PS design (still a noob so forgive me)... is there any advantage to tubes in the power supply to allow a softer "ramp up" to B+ voltages (a softer start-up) than the MOSFETs?

 

[beatnik]

one of the advantages it gives tubes and power supply capacitors an easier life, but in fact soft start can be implemented with solid state regulators too

 

[fazeka]

Thank you beatnik for that explanation.

Another question: I noticed DaveP used two OEP X187B transformers for the control amp input. I'm confused as to how he wired them when comparing his schematic to the original (which has only one control amp input transfomer). So it seems that DaveP's schematic ("T7") shows TWO control amp input transformers... OK.

It seems that the OEP X187B primaries are wired in parallel... and the secondaries of these two are wired in series? Yes?

Thanks all for helping me confirm my understanding! Very appreciative!

Chris

 

[Heikki]

Not blasphemy, thanks for your design! Just to confirm, looks like J4 = control voltage, J7 = signal amp and J8 = 6V6 grids?

Also, it seems the STW8NK80Z is not readily available from the major suppliers, at least in the US. Some saying mid 2023 availability!

J4 = unregulated B+ voltage to control amp. J7 = signal amp B+ and regulated voltage to control amp (12AX7). J8 = 6V6 screen grids.

The mosfet doesn't need to be STW8NK80Z. Almost any N-Channel mosfet with high enough voltage rating should work.

Receiving tubes don't need or benefit from soft start. Direct coupled cathode followers can be damaged on start up if the B+ is present before the cathodes are hot. There is a simpler fix for direct coupled cathode followers than soft start circuit.
https://www.valvewizard.co.uk/dccf.html

 

[fazeka]

You guys see any issues in using a Hammond 812A for the control amp input?

 

[beatnik]

I have been getting inspiration from this thread while planning a Fairchild build and I wanted to share these two options for the control amp output transformer which are very well priced.

https://shop.piemme-elektra.it/Categories/339/Products/8134
https://shop.piemme-elektra.it/Products/7261

They come with two different ratios 4:1 which I believe matches the original Chicago transformer and also a 7:1 option, I was wondering if the higher ratio has a further advantage in achieving a low impedance output from the control amp.

By the way I was wondering if you have been able to figure out the impedance ratio of the Hammond 1645 because I cannot understand what would be the impedance of the 70V tap the way you have used it.

 

[DaveP]

By the way I was wondering if you have been able to figure out the impedance ratio of the Hammond 1645 because I cannot understand what would be the impedance of the 70V tap the way you have used it.

The 1635A is a 10W transformer. 10W/70V = 0.14286A

70V/0.14286A = 490 Ohms

best
DaveP

 

[beatnik]

The 1645 is 30W so that would mean 163ohm right ?

However I am seeing on the Hammond datasheet the 70V connection also puts in series the 4 ohm tap you have used for feedback.

1645 datasheet

If we were to take the output from the black/yellow and white terminal as I believe you did, would this change things or the 4 ohm winding is a negligible difference ?

 

[beatnik]

My only guess is the 7:1 transformer would further reduce the output impedance but the tubes might have a shorter lifespan. Possibly the 4:1 transformer is sufficient and the tubes will have an easier life or there is something else I am overlooking ?

 

[DaveP]

The 1645 is 30W so that would mean 163ohm right ?

Sorry, my mistake. The 70V output was used to many power matching transformers such as would be used to provide multiple speakers in a department store or factory.

I did a lot of experiments to obtain the 100 ohm output impedance necessary for the Fairchild's performance. I used the 4 ohm feedback winding to achieve this. (my feedback resistances are different from the original because of the different transformer) You need this low impedance to charge the timing cap in 100uS.
best
DaveP

 

[DaveP]

The 7:1 transformer would further reduce the output impedance but the tubes might have a shorter lifespan. Possibly the 4:1 transformer is sufficient and the tubes will have an easier life ?

You mean the other way around I think? 7K would be easier on the tubes.
best
DaveP

 

[beatnik]

Yes my bad.

I am actually planning to use 6005W tubes and I'm trying to figure out which of the two transformer options would be better.

I am leaning towards the 7:1 alternative purely because these tubes have a lower maximum plate voltage rating (which will be accommodated by a different PSU) so at a plate voltage of 250V the plate to plate load resistance on the datasheet is 10K, whereas on the original the 6V6 and 6973 are operating around 400V so the plate load resistance is reduced. I am guessing the 4:1 that is originally meant for 6V6 and 6973 would not be ideal for the 6005 tubes. What do you think ?

 

[DaveP]

As you say, the spec sheet for the 6005 states 250V with a 10k push-pull load. This tube is on the limit for voltage/current already for the job it has to do, so 7k would not be a good idea from my point of view. An EL84 or 6HB6 would be a better option because these tubes use 8k loads.
best
DaveP

 

[beatnik]

I also thought about the EL84 initially but 6005W are still available in NOS, military grade versions for very good prices. And with max 12W power I thought it could do the job.
There are Fairchild clones currently on the marked using a ECC99 dual triode output stage, and they sound pretty convincing.

I have also seen more than one design pushing these 6005W tubes in excess of 300V as long as within maximum dissipation ratings, and I was expecting the feedback from the tertiary winding to further help lowering the output impedance.

With the EL84 would you still keep the original 4:1 transformer ?

 

[DaveP]

With the EL84 would you still keep the original 4:1 transformer ?

No it needs 8k
I have attached an article by Merlin which helps you choose the correct Transformer ratio.
best
DaveP

 

[beatnik]

Thanks for the link, will definitely do some more studying of this theory concepts.
I yet don't understand why you say the 4:1 wouldn't work with the EL84, when the 6973 typical operation in fixed bias also expect 8K as on the datasheet and in the original 670 it's used with the 4:1 transformer
My uneducated guess is that the negative feedback from the tertiary helps lowering the impedance ?

Anyways I guess I will have to rule out the 6AQ5 idea because even if it could be adapted to work it would really be maxing out its ratings, which would result in too much heat and short lifespan. So probably not a wise choice particularly for a PCB build with the tubes mounted inside the enclosure

I guess the EL84 as you suggested would be a better choice or perhaps even the EL86 which has a lower plate impedance and increased power output capability, what do you think ?

 

[DaveP]

When you read the article, you will see that the tubes operate in both class A and class B at different parts of the cycle. The load that the tube sees varies between half and a quarter of the load between the plates. So an 8K load will present 4k and 2k which are steep angles on the chart. It follows that a 4k load would present 2k and 1k loads which are even steeper angles. Steep angles can therefore stray over into the overload region much sooner.
Its much clearer in the article.
best
DaveP