I'm contemplating building a 300B PSE amplifier. The Tubelab SE design looks really good, especially with the solid state hybrid setup.
The original design is only for one output tube. In order to parallel the output tube, aside from doubling the power supply current and making sure the output transformer is within rating (2.5kohm load ok?), do any other component values need to be changed?
Thanks
The original design is only for one output tube. In order to parallel the output tube, aside from doubling the power supply current and making sure the output transformer is within rating (2.5kohm load ok?), do any other component values need to be changed?
Thanks
300B PSE's that I have seen use 1k5 or so output transformers, though this isn't set in stone.
I am confused though at what you are doing. I fully understand what you are trying to accomplish, however you cannot just change some component values and some iron and run this amp PSE. The Tubelab SE board is setup for stereo, so at the very least you will need two boards. You will also need to do some surgery (not recommend) to bridge some traces to make it PSE. Why not just find a suitable PSE schematic and build it instead?
Cheers
James
I am confused though at what you are doing. I fully understand what you are trying to accomplish, however you cannot just change some component values and some iron and run this amp PSE. The Tubelab SE board is setup for stereo, so at the very least you will need two boards. You will also need to do some surgery (not recommend) to bridge some traces to make it PSE. Why not just find a suitable PSE schematic and build it instead?
Cheers
James
If needed, I can air wire another tube socket to the existing board. I haven't exactly decided whether or not to stick to a PCB.
I've looked around at PSE designs and I've more or less settled on this one. Andrea Ciuffoli's design looks good but I'm not so keen on using an expensive interstage transformer nor the extensive choke filtering in the power supply.
A design like his (Andrea's) wouldn't be that much more, if not close to the same price to implement. The Lundahl that he uses is about $115, an IT from Electra-Print (which I would venture to say is equal quality) is about $25 less. You would be eliminating the CCS, FET follower, a high quality coupling cap ($$$), and a hand full of passives... I would venture to guess (if a very nice coupling cap is used) close to that $90 give or take. Obviously if you are using orange drops and generic passives then the Tubelab wins hands down.
Either one is a great design, though there is something that draws me towards the Tubelab. Also you don't need that complicated and expensive of a PSU. A CLC would be fine with a RC if you need to drop more voltage and an RC for the front end. Chokes from Electra-Print are about $34 and Hammonds are about $20 or so, it wouldn't be that expensive if you wanted to go CLCLC.
Personally if the shoes where on my feet: 5842 > Electra-Print IT > Parallel 300B's. Stick an LED or string of LED's on the 5842 cathode for bias (eliminate the k bypass cap) and use fixed bias through the secondary of the IT for the 300B's (again eliminate a k bypass cap and a resistor). Use a C-L-C power supply, nailing the B+ by using a custom power transformer (Electra-Print) or by tweaking the first cap, with a pair of 6AX4GTA's for the rectifier. Instead of an additional R-C section for the driver I would use a CCS loaded shunt regulator, CCS made from depletion mode FET's and the shunt from either FET's or gas regulator tubes.
If you stick with the Tubelab design I would make sure to get in touch with George about what you want to do. Not that it wouldn't be possible, I just think it is more work than needed. The Tubelab design can be done point to point too so you don't have to cut up traces and hard wire on a PCB. Also George has a good idea about the limitations of his designs, make sure the board can drive parallel 300B's OK. Just my 2 cents
Cheers
James
Either one is a great design, though there is something that draws me towards the Tubelab. Also you don't need that complicated and expensive of a PSU. A CLC would be fine with a RC if you need to drop more voltage and an RC for the front end. Chokes from Electra-Print are about $34 and Hammonds are about $20 or so, it wouldn't be that expensive if you wanted to go CLCLC.
Personally if the shoes where on my feet: 5842 > Electra-Print IT > Parallel 300B's. Stick an LED or string of LED's on the 5842 cathode for bias (eliminate the k bypass cap) and use fixed bias through the secondary of the IT for the 300B's (again eliminate a k bypass cap and a resistor). Use a C-L-C power supply, nailing the B+ by using a custom power transformer (Electra-Print) or by tweaking the first cap, with a pair of 6AX4GTA's for the rectifier. Instead of an additional R-C section for the driver I would use a CCS loaded shunt regulator, CCS made from depletion mode FET's and the shunt from either FET's or gas regulator tubes.
If you stick with the Tubelab design I would make sure to get in touch with George about what you want to do. Not that it wouldn't be possible, I just think it is more work than needed. The Tubelab design can be done point to point too so you don't have to cut up traces and hard wire on a PCB. Also George has a good idea about the limitations of his designs, make sure the board can drive parallel 300B's OK. Just my 2 cents
Cheers
James
Well, I was going to use the LM-9907P from VT4C.com and probably some choke from there. So the prices should be lower. I already have teflon coupling caps, so no additional costs there.
Two interstage transformers just can't compare to two mosfets (or 4? one per 300B?)
I agree about the power supply and may try the LED biasing too.
Two interstage transformers just can't compare to two mosfets (or 4? one per 300B?)
I agree about the power supply and may try the LED biasing too.
In cost or in performance, because if you are talking cost then truth.
Teflon caps, there is a huge part of your cost right there then that you already have, makes a stronger case for the Tubelab or similar design!
I was looking back over the Tubelab SE schematic and the board, then I used my brain a little bit. The very easiest way to do what you want while using a single board would be to hard wire a second socket to the board (1 board 4 sockets), change the output transformer, and diddle with the bias a little bit. The Tubelab uses fixed bias so you wouldn't have to change any parts, just play with the trim pot that adjusts it. The biggest worries I would have here would be power supply ability and drive. I think drive ability would be OK, George would be a better judge of this. You may be taxing the PSU section a bit by running 4 300B's plus drivers off of one 5AR4.
I think the easiest way to do this would be to build it point to point and use veroboard to build your CCS's and FET followers. Unfortunately it doesn't send any financial support George's way and he is a great guy; I feel every penny he makes from the SE's and Simple SE's is more than well deserved. However it allows for a bit more design flexibility.
Some tubes like LED's on the cathode and some don't, word on the street (at least the street I live on) is that the 5842 is one that plays nice with LED's. I have been using them (LED's) recently as I play around with a CCS loaded 264B as a line amp, the jury is still out on that one.
I didn't mean to jump you so quickly about slicing and dicing on the Tubelab board (though I still don't know that I can recommend it). It just isn't always a good idea to take a board designed for one thing and rig it up to do another. Though after looking back, aside from a few issues (the ones mentioned), it doesn't seem like it would be all that hard.
Cheers
James
PS. George or anyone else... if the FET follower is used does it benefit at all from a CCS load verses a large wattage resistor like the Power Drive has? Not to hijack this thread, just interested as Mazuki may want to know too if he is using it.
I don't think the 5AR4 will cut it, so I was planning to use SS rectification with either a standby switch setup or a cheap delay kit off eBay.
I'll most likely buy the board and then decide to use it or go with point to point. I hope I'm not offending George here by modifying his design for more power.
Either way, it wouldn't be hard to switch between LEDs and resistors and see what sound I prefer
I would go with the "human powered filament delay" AKA a switch, this is what AudioTropic and I am sure others use. It is cheap, easy, and there are no worries about failure unless you have chubby fingers and you put the B+ and filament switches to close together
Cheers
James
edit: Just out of interest, why parallel, what are you driving?
It wouldn't offend me one bit. I have hacked up quite a few of them in the name of sound. In fact many of my tube testing experiments were done by wiring the output tube of board. I have used them to drive a bunch of different tubes from 45's to big sweep tubes.
The first problem I see is driving two parallel tubes with fixed bias. Just wiring two tubes in parallel won't work unless they are perfectly matched. Even if you find two pairs of matched 300B's, they won't stay matched over time. The solution is to use two mosfets, one for each tube allowing individual bias adjustments.
Next, as noted, the 5AR4 wouldn't like feeding 4 tubes. I like to have the negative voltage come up first to avoid surprizing the output tubes. Using a solid state rectifier can allow the B+ to come up quicker then the negative voltage.
LED's are easy on the Tubelab SE and the Simple SE board. Just leave the resistors out and put the LED where the bypass cap goes. The issue stems from the huge unit to unit variation on the 5842. I went through a bunch of different LED's to get the plate voltage right, and wound up with two different color LED's. If you have a bunch of tubes you can match the tube to the LED.
I think that there are two possibilities. One is to build the amp completely point to point, or maybe build a small daughter board for the extra mosfet circuitry.
I can propose another solution. Build the board according to the construction manual. Remove one of the 5842's and wire the two coupling capacitors to the remaining tube. Wire the plates of the two 300B's together and to a 1.6 K or so OPT. This allows one 5842 to drive both 300B's with independent bias. This results in a monoblock of course, so two would be needed. Two power transformers too.
I am planning to hack up a perfectly good working Tubelab SE to make a tube evaluation test fixture. I can test this concept if there is any interest. It will be 2 to 3 weeks before it will be ready.
tubelab.com said:
The other really cheap option is to entirely fake the matching.
Abusing one real 300B as the reference model. Then cloning
a perfect copy out of a stronger substitute device.
Anti-triode SEPP comes to mind, which also solves the problem
of dirt cheap OPT. You could abuse MOSFet, IGBT, or Pentode
as the opposing Anti-300B. Its as simple as Class A push pull,
with a huge deliberate mismatch of Gm, and the Antiside ref'd
to GND.
Its stronger side is pretty much forced to conduct exactly the
opposite of the weaker side (the 300B). And the net effect is
nearly the same as two identical SE 300B's working in parallel.
bigwill said:
AC current in the Anti-side not constant. AC currents are
equal and opposite. DC currents are equal and the same.
DC currents in both branches might be held to an identical
pair of constants, but its certainly not the only way.
Its like a cascode turned upside down. AC current off the
300B cathode goes right back up to the other side of the
transformer through the Anti-Triode. But all 300B cathode
"sees" is a virtual ground, held constant by the huge Gm
of the Anti-triode, which is merely following GND.
The Anti-Triode is also usually a device with no sensitivity
to plate voltage, IE: the plate impedance extremely high.
Thus the current is always slave to cathode of the 300B,
SE and not flavored by any properties of the Anti-device.
----------------------------------------------------
Paraphase, driven from a voltage divider between plates,
with a grossly overmatched slave device, would also grant
the equal and opposite SE effect. Yet another cheap and
dirty way one might implement an Anti-triode clone.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.