Poseidon motivation... or, Dynaco upgrades 'r' us.

[anarchist]

There was a thread last month that conveyed a great deal of tube knowledge and centered on Dynaco upgrade possibilities. Some members here have been sitting on boards and/or their projects for a while. I thought I might share my progress and motivate them as well.

I purchased the Poseidon boards. The fellow who makes these, Shannon, is a great guy and willingly discussed aspects of their design and motivation behind their design. I also conversed quite a bit with Uncle Ned about his modifications and reasoning.

The boards arrived in an envelope inside a larger box with the instructions and the auricaps I purchased. This circuit board thing is just too cool. Point to point had been my preferred topology but circuit boards make things too damn easy. Attractive, good looking thick boards which are layed out nicely and marked on both sides. The instructions are largely the parts list and schematics. Ned offers more detailed instructions and I suggest downloading them.

Ordered all the parts from mouser and substituted a couple of the stock resistors with different values (based on Ned) to obtain slightly higher plate voltages. The differences are rather subtle in the voltages and either would work. The theory is the higher plate voltages would result in less distortion. Stock these boards use a 12ax7 and a 12au7. You can substitute a 5751 and a 12bh7 in the circuit if you wish with no changes. As mentioned, in my case, I chose to increase the plate voltages a bit.

Boards assemble pretty easy. I didn't spend much time at all putting them together. Its simply a matter of bending the wire, taping a series of parts to the board, and then soldering them at one time. Or you can do it one part at a time if you want to go slower. On my boards, the auricaps will be placed on the bottom side of the board you see. The plan is to recess the board in the Mark III's and then install shaped plexiglass covers over the circuit boards with holes providing access to the tube sockets. I think the look will be great, match the modern look of my Mark III's and the boards protected from curious fingers.

[3dzapper]

C'mon Robert, let those fingers in there! You're no fun.

Looking good. I've decided to get a pair to try in my clones. Just waiting for a change in the billing cycle if you know what I mean. I have to post a question I have to Shannon's forum.

Rick

[anarchist]

Hi Rick,

Glad you are grabbing a pair. It is fun and these boards have received some good reviews. I also know there are various modifications you can make to them to change things up more including some mods which involve cutting some traces, etc. than Shannon will help with.

Hi Mark,

Therein lies the rub. Currently I have no preamp so I haven't even listened to these things yet. Still getting some of the pieces in place: I have some NOS Mullard 5ar4's on order, I still need to place orders for the 12bh7 and 5751. Before actually doing anything with the amps, I want to collect some measurement data with the stock cans and 5u4 in place and do some listening for a week or so. Then I think I will order some thermistors and upgraded cans and stick in there at the same times I swap the boards and tubes, collect some more measurement data and do some critical listening. I plan to fully know the nuances of these amps before I done and be able to translate that to future projects.

Ah, the bias circuit. I still need to become more familiar with how the bias circuit is affected by the rebuild Will performed and the change to these new boards. My basic understanding right now is the amps are currently biased with the stock rotary dial. I am not certain how the poseidons affect that. It seems you still use that bias control but now you also bias on the board itself. I am still trying to understand - probably be easier with everything in place and the DMM in my hands - these instructions:

1) Remove your 5AR4 and power the amp. The bias voltage will appear
immediately at J6 and J21 on the diytube PCB. Adjust the control on the amplifier
chassis so that these voltages go as negative as possible (eg -55VDC), but
adjust to have the voltage the same on both test points. You can use the preamp
socket pins 2 or 3 as your DMM ground.

2) Turn the unit off. Replace your 5AR4 rectifier. Turn on unit.

3) Because adjusting one side usually changes the bias on the other side, you
will now start ping-ponging between the sides adjusting the wire wound pot and
measuring one of the bias points on that side unit you near 500mV.

4) Now that you are around 500mV on all the tubes, stick your DMM probes in
pins 5 & 8 of the preamp socket. You are measuring the small voltage difference
between the biasing of the 6550 or EL34 pair. Now adjust the R15 balance
control until the voltage you read is zero or near zero difference. You might end
up changing the scale of your DMM to the 300mV region.

5) Now repeat step #3, but adjust chassis control to produce to 700mV (for 6550
or KT88) or 600 mv (for EL34) across the 10 ohm resistor.

I, of course, am open to modifying the bias circuit and improving it. I know improvements can be had by increasing the filtering (unless Will has already done so) and I would like to implement a true individual tube bias for the KT88's.

[Marvel]

Robert,

The original circuit sets the bias for both output tubes simultaneously. Shannon's boards allow your to get the bias for each tube set correctly. This means it is not as critical to have matched output tubes.

Bruce

[fini]

Robert,

Thanks for doing all this research. Hopefully, most of it I can apply to my MK IVs. I'm off the rest of the week, but so is my wife (which means I've got chores to perform that do not include rebuilding tube amps)...

[anarchist]

Robert,

The original circuit sets the bias for both output tubes simultaneously. Shannon's boards allow your to get the bias for each tube set correctly. This means it is not as critical to have matched output tubes.

Bruce

Thanks Bruce. I didn't explain my uncertainty that well in my last post. I had gathered it was trying to solve the matching bias issue and Shannons boards provided a workaround for the individual bias pots (basically shifting power from one to the other although I don't quite understand how he is doing it) but I guess I question if having individual bias pots would be better. I need to study the schematic a bit more. The possibility of then eliminating the Poseidon's additional bias circuitry from the equation then becomes my educational experience and target. I want to understand the improvements the differing approaches would make.

[anarchist]

Ok. Studied the schematic a bit. I haven't done the math and while I can I don't know what it would tell me yet. Basically, the new circuit board sticks a 500K pot before the original bias circuit to adjust the voltage going to the 2 100K resistors which are then adjusted by the 10K pot. Reviewing the poseidon schematics in detail shows they by design suggest further modding if the bias circuit. I have created a small picture to show the changes from the original. Basically, they double capacitance and change the 1K resistor in the circuit to a 4.7K. Simply put they have increased filtering. I know from observing the selenium rectifier in the circuit would do well to be replaced a diode I believe. How am I doing? Do I have all this right? Could this be improved upon further? Given these changes, is there really a need to put in a second bias pot to control the tubes bias independently or does the 500K pot do the deed well enough? Its starting to get fun.

[3dzapper]

On my amps, the 100k resistors rather than fixed, are pots tapped at the output of the 1K resistor. Each tube can then be biased individually.

Hey! Want a nice tube tutorial? http://www.du.edu/~etuttle/electron/elect27.htm#PowTri

Rick

[anarchist]

Ok. Figured it would be a good idea to know exactly what was in my amps after their rebuild in order to determine what bias improvements there were to be made. Discovered the bias circuit in these boxes has been changed. The selenium rectifier is gone and replaced with a diode. The caps have been upgrade to xicon 100mfd/100v with 20% tolerance. The pot has been changed for 10k to 25k. I can't identify the resistor yet (it could be a 1.6 or a 7.6 hell if I know - have to find my DMM) and the final piece is the .02 disk cap.

Edit: Check with my meter and it looks like the resistor connected to the diode is a 1K. The resistor on the pot connected to ground is fun. It should be the 18K but appears to be marked a 7.6K and it is hard to tell and when I place the meter on it... it climbs rapidly and then slowly to 9.94K. I am assuming because it connects to a capacitor I can't get a reading of the resistor itself and I am witnessing the cap charging?

[anarchist]

Yes. These were Dee's amps and were rebuilt by Will. He altered the layout a bit but they are stock boards, chokes, etc... but I know he did modify the bias circuit after looking at it.

[3dzapper]

"Edit: Check with my meter and it looks like the resistor connected to the diode is a 1K. The resistor on the pot connected to ground is fun. It should be the 18K but appears to be marked a 7.6K and it is hard to tell and when I place the meter on it... it climbs rapidly and then slowly to 9.94K. I am assuming because it connects to a capacitor I can't get a reading of the resistor itself and I am witnessing the cap charging?"

The original circuit from the diode to ground was 29KOhms, With the change to a 25K pot plus 7.6K and the 1K is 33.6KOhms. You should be OK. More of the voltage will drop across the pot, that should give more adjustability.

Rick

[anarchist]

Yep. I figured as much. With the assistance you guys gave me in the big thread last month, I am doing pretty good I think. I have been able to reconcile everything I am seeing in this amp with the schematic. If there were some room in these things, this would be pretty damn easy to do.

I guess my only real issue now with dropping the Poseidon in is I need to eliminate the 11.2K resistor from Pin 1 on the 6550 as is on the original circuit and replace that with separate 10K resistors on each 6550 to ground. Ok that is easy enough.

But... the original circuit connected a jumper from pin 8 on V1 to pin 8 on V2 and then wired it to pin 8 on the Preamp socket. On this one there is no preamp socket. The two Pin 8's are both directly wired to a pin jack. OK. That seems to me to result in the same thing.

However, the bias instructions with Shannon's board tell you to place a probe in Pin 8 of the Preamp socket (easy enough, use the pin jack) and place the other probe in Pin 5 of the preamp socket (uh, no such thing) and you are measuring the difference between the two tubes. Checking, I find that Pin 5 was connected to was a 22K resistor and then terminal 5 on the original board. So then...

Not exactly certain how that works to tell the difference - seems Pin 5 is negative voltage and Pin 8 positive so there is the difference - but in the original circuit you tested via Pin 8 and Ground. I have two Pin jacks that provide Pin 8 and Ground. Do I need to rewire the "Ground" pin so it actually is now terminal 5 (Easy enough to do)?

My understanding of the issue was if you set the pot at 700mv, one tube might be 700mv and the other 650mv, so doing away with this Pin 8 and Pin 5 business, couldn't I just run individual wires from Pin 8 on each socket to the two pin jacks. I would then know exactly what each tube was biased at and could use the 500K pot on the Poseidon to "balance" them... just trying to reconcile this bit and do it the best way. Thoughts?

[fini]

Robert---

Yes, when you place your meter on the 18K you'd be measuring 3 paths including a couple with capacitors! You'd have to disconnect it. You'll find that this is often true when attempting to measure parts in-circuit.

This is a great point, actually. I have a question about unsoldering parts to either check their value, or to swap them out. Do you "old pros" consider this no big deal, you can do it as many times as you want? It seems like it would be a time-consuming process, epecially if leads are, say, bent through lugs, or if there are several parts soldered to the same point. Are these concerns dumb?

[3dzapper]

GRRRR!!! The internet ate my post!!!!

Those instructions you cited are to set the original bias into a safe range for final bias adjustment. Once it is done, there is little likelyhood that it will ever be repeated.

The final bias is measured from pin 8 to ground across the 10 Ohm resistor. There are instructions on how to safely measure final bias on the octal at the back without removing the base in Uncle Neds Poseidon manual on page 8 step 7.

(download here: http://www.diytube.com/unidriver/poseidon-tz.pdf )

Rick

[anarchist]

Thank you Mark. Quick reading huh, whats a "K" here or there. LOL.

4) Now that you are around 500mV on all the tubes, stick your DMM probes in
pins 5 & 8 of the preamp socket. You are measuring the small voltage difference
between the biasing of the 6550 or EL34 pair. Now adjust the R15 balance
control until the voltage you read is zero or near zero difference. You might end
up changing the scale of your DMM to the 300mV region.

5) Now repeat step #3, but adjust chassis control to produce to 700mV (for 6550
or KT88) or 600 mv (for EL34) across the 10 ohm resistor.

Here are the relevant parts which have me confused. I don't have a preamp socket so there is no pin 5 or pin 8 for that matter. Pin 8 on the tube sockets are connected in parallel to a Pin jack. A second pin jack is connected to ground. As is, this would allow me to perform step 5 - final biasing, right?

But I need to balance the bias between the two tubes in step 4. So... if I change each pin jack so that each is connected individually to pin 8 on the 6550's and use some other point as ground, I am looking at the final bias of each tube individually and can adjust the 500k pot so they are equal no?

Mark seems to be saying if both those pin jacks are connected to pin 8. Then I can simply stick one probe in one and the other probe in the other and that will also tell me the difference between the two.

I am sorry for the dumb questions but this part has me confused as hell. If I could turn on the amps and start measuring it would probably make sense quickly and I could figure out how to accomplish this.

[anarchist]

Whoo hoo. I ain't so stupid after all.

Now let me thank you for helping me sort through this and say I can appreciate how hard it must be to grasp what I am asking or describing and provide assistance.

Now let me be a moron and ask a dumb question. [^o)]

The pin jack that is currently ground. (Can't wait to try and explain this)

It would seem simple enough to just desolder the connection to it and solder the jumper from the other 6550 tube socket. But... it has a grounding bar that runs to a lug terminal strip which is then bolted to the chassis. The 100mfd cap that connects to the chassis pot ties to this grounding bar. The 18K resistor in the bias circuit is attached to that same grounding lug.

The question is actually pretty simple and I am pretty sure after trying to explain it I know the answer. The ground lug is providing the ground to that bar and in turn the capacitor and the pin jack. I can clip off the grounding bar at the pin jack with no worries. Right?

[anarchist]

Hoorah! Those lessons have paid off. I was even able to reconcile the differences between an original dynaco, a modifed Dynaco with a different layout and bias setup, and the modified dynaco with the Poseidon board. With some assistance, of course. Can't run before walking but this stuff is getting simpler to grasp and tie together. Combine theory with some hands on and it becomes real knowledge.

Thank you guys again.

Feel free to tell me what I missed and where I am going to get myself in trouble. [:$]

[3dzapper]

OK I see your confusion Robert. The "Pre-amp socket" Ned is refering to is the octal socket on the back of the chassis. In addition to being a bias test point, that socket was designed to provide power to some of the pre-amps of the time.

I typed this a while ago. Then we went to Lowes. Senior moment, I forgot to hit post.-(

One thing I don't see in the manuals is a hook up between the KT's chassis and the test/pre socket. In the original Mark III manual and build, both KTs combined bias is measured at pin 8.

Make sure that the wire from one KT's pin 1/8 to the other is removed, the bias measurement will then be taken across each tube's 10 Ohm cathode resistor. To measure bias from the back test/pre socket, a piece of hook up wire must be run from pin 1/8 of each KT. One would go to pin 8 of the test socket, the other to pin 5. Those biasing instructions then begin to make sense.