• 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.

OPUS 5.0 A Modern Mullard

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2003
Paid Member
I was looking for a good project to expand my knowledge and fill a need I have with my Audio System. I desired an amp to drive my base drivers on my 3-way using an active cross over (very timely SY and Pmillett ). Presently the Speakers are being driven by a TubelabSE 300b with a passive xovr. They are loud enough, but often as expected, bass notes challenge it beyond its capacity.

The goals for the amp were around 60W, with decent dampening factor for bass duty. To do that I knew I had to go to a PP design, which I wasn't excited about initially. I had considered the Williamson but quickly turned my attention to the Mullard based on feedback on this forum.

After reading Tubemacks thread with the ideas and discussion within it, I was very impressed and excited. The attached schematic is a derivative of that thread. Out of respect for Tubemack and his goals, I started this thread so it would not derail his goals.

I would like to thank everyone that has helped me so far, particularly Eli, SY, boywonder, and M6tt. I am sure I am missing others so please forgive me.

Output Tubes kt88 in UL
LTP is the 12at7 (This to me seems like a good answer based on Eli's experience to the listenability issues I had with commercial PP's I have heard)
The Input driver at this point is the 12bh7 per Eli's suggestion which would fit the drive requirements of the t7 ltp and not require TOO much feedback to tame the gain

With the 12bh7 about 15-16dB of feedback should be sufficient.

I have choosen the powerdrive from tubelab to provide a low impedance drive to the kt88 and provide a high impedance load to the t7 ltp. Even though it has AB2 capability it should stay within AB1 operation with very few exceptions. The advantage of this, in my mind, is that any grid current will be handled with authority.

If you would, please evaluate the schematic and comment freely on it.

If the operating voltages and currents look close enough, I will design and post the PS schematic. My hope is to use a PT with a single primary, bias, and possibly rectifier filament winding. It will be SS rectified at this point in design, but if I have a PT wound with a rectifier winding it has a higher recycle value in my inventory for other projects should I choose. The filaments will be heated with a seperate xfmr to allow them to be heated before application of B+.

Thanks Again
Scott
 

Attachments

  • Visio-ab2schema_v5.pdf
    43.8 KB · Views: 2,904
Member
Joined 2003
Paid Member
The FET followers could easily be switched for a 4k R for a little less sand. I left them in as they will sit right next to the FETs on the heatsink.

I am running the kt88 at 475V which is hotter than 465V in the tubemacks thread. This gives the t7 ltp a lttle more room (not much mind you) to et it set up right. The higher impedance gives IMHO a better loadline at 475V. That is ONLY based on my limited experience.

The GEC tubes came in yesterday. I want to get opinions on this and post the PS schematic for review before I order the iron. I only want to do that once!!!!
I have also sent Bud at O-netics and Jack at electraprint an email but havn't heard back from them.
 
Last edited:
R1 is too big. 100 KOhms is about right. R1 combines with V1's CMiller to form a low pass filter. When R1 is too big, HF info. gets lost.

GNFB should be taken from the speaker tap actually in use. Got 16 Ω drivers? If you're going to have Jack E. or Bud P. wind O/P trafos, think hard about a single O/P tap of 6 Ω. I'll let somebody else check your math. I foul it up far too often.

R9 and C2 form a high pass pole. For stability's sake, F3 of that pole should be as close to zero as can be, consistent with practical considerations. Also, C2 needs to be of very good quality. With those points in mind, a 400 WVDC/680 nF. PPFX MultiCap appeals to me. Michael Percy has them at 9 bucks a pop.

The R19/R20 gate stoppers are too small. The transconductance (gfs) of power MOSFETs is high. Use 510 or 1000 Ω Carbon comp. parts here.

A cap. is needed at the I/P, which will form a high pass filter in combination with R1. F3 of this pole should be in the 15 to 17 Hz. range. A properly sized high pass pole here protects against O/P trafo core saturation from a huge GNFB LF error correction signal. A Soviet surplus PIO part is dandy here, as this pole is outside of the GNFB loop.
 
Member
Joined 2003
Paid Member
Revised Schematic per suggestions

Okay, Revised.
Almost all my SE OPT's are 6ohm, so no problem there.
Little hesitant to put another cap in but point is valid so in it is. Should be right around 15Hz. Noted C2 and set value. 510 CC in for gate stoppers.

Revised Schema Attached.

Those with experience with kt88 in PP, does the approx OPT seem reasonable for operating point? Or is Operating point too aggressive? I don't need the tubes to last until I turn gray, but I don't want a melt down either. (wait too late for the gray).
 

Attachments

  • Visio-ab2schema_v5_1.pdf
    43.8 KB · Views: 854
Member
Joined 2003
Paid Member
Power Supply and revisions

Here is the power supply and Updated Driver
B+ has 0.8mV ripple at the kt88 The ltp has 0.12mV ripple and the driver 0.08mV

C- has 1.2mV ripple

The filaments are switched on first which enales the HT to be switched on.
 

Attachments

  • Visio-Opus5_0 Power Supply_v1.pdf
    41.9 KB · Views: 1,073
  • Visio-Opus5_0_Amp v5.pdf
    44.2 KB · Views: 736
Some PSU comments.

C110 is WAY too big. If you want the 7805 IC to be reliable, C110 should be 22 μF. Put the 1000 μF. part after the rectifier bridge, where it will do some good. Instead of nasty Tantalum, I suggest a Panasonic ECQ-V stacked film part as C109, as close to the 7805 as can be. Why a 5 VDC rail, to begin with?

The signal will ride on the negative rail. LOW ripple is important. So, make C111 much larger and install a "hash" filter here too. Definitely use Schottky diodes in the negative rail bridge. Quiet is very good here.

Install ferrite beads on the leads of the filament trafo's primary. Keep crud from sneaking into the signal, via the heaters.

Have you given thought to B+ rectifier diodes selection?
 
I'm assuming the PS transformer will be a custom, correct?....and C113 & C117 are motor runs....

Nice heater safety-interlock human powered delay, have you thought about using a amperite-based delay? There is a pretty simple schematic around that uses an amperite and a DPDT relay to do the time delay, then takes the amperite out of the circuit so it can cool down for the next power cycle. I have a copy somewhere if you are interested.
 
Member
Joined 2003
Paid Member
Eli, I am not sure I really want to use the 7805. Actually was thinking of the lm317 when I used the fixed device. Of all things, I missed the filtercap prior to the 7805. Fixed. The rail os for some digital panel meters so I could see bias. I thought of putting in just a test point.

Hash Filter goes in next version of bias supply drawing. Oh BTW I had a typo, unlike what was posted above, the simulated ripple is 0.133mV on the bias. I will increase the caps a little, but I run into a time issue. With the large dcr of the choke, a bigger cap slows the establishment of the bias. In my present design, that isn't big problem as I have the huge caps with the thermistor (not shown yet, oops) delaying b+. If more normal size caps are used inn the B+ circuit, the the b+ WILL come on alot quicker.

As far as diodes, will need something >1200 V. PSUDII is showing just under 1100V PIV. I was too tired to spec those parts last night. I got as far as developing the parts list for the amp section and then I couldn't read the fine print in the mouser catalog ;-)
 
As far as diodes, will need something >1200 V. PSUDII is showing just under 1100V PIV.

That's real easy. Use series wired pairs of UF5408s or even UF4007s. Snub each pair with a 10 nF. part of the highest WVDC you can get/afford. Even though a cap., by itself, is an imperfect snubbing setup, it's quite adequate. The "hash" filter keeps switching crud out of the B+ rail and there are no filament windings on this power trafo core.
 
Member
Joined 2003
Paid Member
PSU Revised

Latest PSU attched. I'll check the UF5408s, I put the snubber caps in as it is a good design practice.

Schotky's for the other locations.

boywonder, I am interested in the delay setup.



FYI, Bias ripple now at 0.043mV
 

Attachments

  • Visio-Opus5_0 Power Supply_v2.pdf
    42.7 KB · Views: 420
Last edited:
Member
Joined 2003
Paid Member
I am running a little hotter b+. 75mA pushed the kt88 too close to max plate power. It is a simple tweek at this point to change however, the transformer will be speced at a little higher current and then some current will be sunk to bring the supply in line. That will give me flexibility to run it at different bias points. (NOTE: I just realized I didn't take into account the current to be sunk from the FET Followers)

I have NO experience with PP designs. So I am absolutley ALL ears on he best way to set it up. I choose the higher voltage since I am not placing a limit on b+ to fit electrolytics. The higher B+, does give a little more room to set up the ltp. (not much though).

I attached what I was thinking
 

Attachments

  • Visio-Opus5_0 Power Supply_v3.pdf
    43.1 KB · Views: 255
Last edited:
Latest PSU attched. I'll check the UF5408s, I put the snubber caps in as it is a good design practice.

Schotky's for the other locations.

boywonder, I am interested in the delay setup.



FYI, Bias ripple now at 0.043mV


Scott,

Have you given thought to selection regarding C117, which bypasses the gargantuan 'lytics? Something that "lays the lumber" to ESR and ESL is in order. With that in mind, I suggest a 630 WVDC/40 μF. Axon MPP bypassed, in turn, by a 600 WVDC/1.0 μF. PPMFX MutiCap. Respectively, Michael Percy's prices for those parts are $15.95 and $11.50. Obviously, shopping around is appropriate.
 
Member
Joined 2003
Paid Member
I was going to use a GE Motor Run there as well. I have a quite a few laying around from a purchase a while back from a surplus house.

I was going to put a poly film on top of that. maybe a ppmfx.

Let me make this statement to those following along. You don't need the monster caps in this thing. I just happen to have 4 x 3500uF 450VDC caps staring at me from an industrial DC Drive. And Yes I store them with ground straps. There is enough energy in these things to REQUIRE an "ARC FLASH" study and to make a enclosure to contain it. Because of my line of work I am used to these requirements.
(Although steel mills don't use audio tube amps ;-) ).

If enough interest, I would be happy to model a more "reasonable" Power Supply.
 
I was going to use a GE Motor Run there as well. I have a quite a few laying around from a purchase a while back from a surplus house.

I was going to put a poly film on top of that. maybe a ppmfx.

Let me make this statement to those following along. You don't need the monster caps in this thing. I just happen to have 4 x 3500uF 450VDC caps staring at me from an industrial DC Drive. And Yes I store them with ground straps. There is enough energy in these things to REQUIRE an "ARC FLASH" study and to make a enclosure to contain it. Because of my line of work I am used to these requirements.
(Although steel mills don't use audio tube amps ;-) ).

If enough interest, I would be happy to model a more "reasonable" Power Supply.


GE motor runs are FINE! So, the monster 'lytics are for an industrial DC drive. That suggests good heat tolerance and a long service life. No problems in this PSU with current reserves for transients. :D When you measure the finished product, I expect 3+ dB. of dynamic headroom will be available.
 
Member
Joined 2003
Paid Member
Looking at the HT diodes, I would like to stay with a 1200V part.
Two come to mind
1) IXYS DSEP30-12A-ND (I think Tubelab had problems with the new IXYS)
2) Fairchild RHRP8120. 1200V

Also what would be a good ferrite bead? The list is endless and I am not sure how to spec it.

As a final note I am omitting the 5V supply and meter and going with test points. Easier to implement and elminates issue of getting noise back into HT and bias supplies.


The moster caps are BHC Ultra Long Life. It is a PERFECT Example of Dumpster Dive. The drive was actually good with minimal usage. It was being tossed cause we switched to AC Freq drive in that application. I also pulled a perfect heat sink from it to which will be used in this project.
 
Last edited:
Looking at the HT diodes, I would like to stay with a 1200V part.
Two come to mind
1) IXYS DSEP30-12A-ND (I think Tubelab had problems with the new IXYS)
2) Fairchild RHRP8120. 1200V

Stacked and snubbed UFnnnn pairs are low cost and work well. In 1200 PIV parts, DigiKey stock # C2D05120A-ND rules the roost. It's a Silicon carbide (SiC) Schottky made by Cree, inherently quiet, and no snubber cap. is needed. Search the archives for "cockeyed bridge". It's a technique for using a combination of fast PN and Schottky parts, which emphasizes the cost and noise performance of the 2 technologies.

Also what would be a good ferrite bead?

A part that's chamfered, to prevent nicking wire insulation, is appropriate. The central hole should be large enough in diameter to permit at least 1 turn of wire around the bead's body.

As a final note I am omitting the 5V supply and meter and going with test points. Easier to implement and elminates issue of getting noise back into HT and bias supplies.

Good move. ;)
 
Member
Joined 2003
Paid Member
Latest PS schematic. Decided on SiC Schottky. Left snubber caps in for now, I don't think it would hurt. I also updated wattage rating of components to give higher margin of safety.

I agree UFnnnn will work stacked and would be a good fit. For right now easier to think about 1 piece. For $6 diff in price per block it is worth it. I will re-read cockeyed bridge.

Getting close to my goal of having parts list complete to make orders tonight.

I would like to get a little more feedback on trafo's and experience at my indicated operating points before I order them however.
 

Attachments

  • Visio-Opus5_0 Power Supply_v4.pdf
    42.4 KB · Views: 384
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.