|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
98.209.168.176
In Reply to: RE: Garg0yle 2A3 DC posted by Garg0yle on January 19, 2015 at 11:36:40
this is ironic I think.
is there no cap after the choke?
are you using the 5h 135ohm choke?
Follow Ups:
The capacitor from 2A3 cathode to ground should really be 150uF, since is is sitting at 40% of B+. That would give "matching" voltage fluctuations. I stayed with 100uF so that all the caps would be the same, and none of them too big to make film caps practical. (See the Rev 1.1 diagram.)
If you remove that cap entirely, there will be a big loss in filtering, so a cap from B+ to ground would be needed, I think.
Certainly an added stage of filtering would reduce ripple; I didn't do that in order to maintain the necessary high voltage. Substituting a solid-state rectifier would provide some extra voltage, if that is needed. Cheaper than a new power transformer!
Do you have specs (especially inductance) for the Hashimoto transformer?
If the driver plate choke DC resistance is changed, the 2A3 cathode resistors may need to change to maintain the target voltages and currents.
Since I put this up as a community project, I have no problem with revisions. I think it would help everyone if changes are labeled, just so we can track the effect of variations.
"The capacitor from 2A3 cathode to ground should really be 150uF, since is is sitting at 40% of B+. That would give "matching" voltage fluctuations. I stayed with 100uF so that all the caps would be the same, and none of them too big to make film caps practical. (See the Rev 1.1 diagram.)"
Makes sense.
"If you remove that cap entirely, there will be a big loss in filtering, so a cap from B+ to ground would be needed, I think."
I would just assume leave it below the ultrapath since I can balance it out and or increase the total capacitance.
"Certainly an added stage of filtering would reduce ripple; I didn't do that in order to maintain the necessary high voltage."
Actually the ripple is pretty good.(just to clarify I view "ripple" the bit of AC left over from the rectifier and not voltage variations caused be the load.)
So maybe I can get by with just added C.
"Do you have specs (especially inductance) for the Hashimoto transformer?"
13H/100mA on the 2.5K Ohm tap.
I think the primary should be 99 Ohms, I'm not sure if that is the 3.5K tap though.
"Since I put this up as a community project, I have no problem with revisions. I think it would help everyone if changes are labeled, just so we can track the effect of variations."
OK
Thirteen henries is moderately low for a 2.5K transformer; that can give you perceptibly weak bass, especially at high levels. The Hammond 125ESE is spec'd at 10 henries. I like to go for 20 henries at 2500 ohms; beyond that you get big transformers with questionable treble - in my experience at least.
I totally agree that better power supplies make better music. A shunt regulator for this amp would be a beautiful thing. But in series feed you need ore current through the shunt reg than through the amp itself, so the regulator will dissipate at least 30 watts - no longer a simple, inexpensive project!
Incidentally, I did not design this for parallel feed even though I personally prefer it. A lot of people seem uncomfortable with it. But it does provide a good measure of power supply isolation.
"Thirteen henries is moderately low for a 2.5K transformer; that can give you perceptibly weak bass, especially at high levels. The Hammond 125ESE is spec'd at 10 henries. I like to go for 20 henries at 2500 ohms; beyond that you get big transformers with questionable treble - in my experience at least."
The OP transformers do have a 3.5K - 18H @ 80mA taps along with variable 25W cathode resistors.
"I totally agree that better power supplies make better music. A shunt regulator for this amp would be a beautiful thing. But in series feed you need ore current through the shunt reg than through the amp itself, so the regulator will dissipate at least 30 watts - no longer a simple, inexpensive project!"
That's OK, I'm confident something can can be worked out with what you have here. IMO
"Incidentally, I did not design this for parallel feed even though I personally prefer it. A lot of people seem uncomfortable with it. But it does provide a good measure of power supply isolation."
I was somewhat interested in parallel, although I already had these OPTs, so just as well for myself!
What counts is the ratio of inductance to impedance, which is the same on both taps.
"What counts is the ratio of inductance to impedance, which is the same on both taps."
-OK Gotcha.
That's a neat design, somehow I missed it!
My immediate reaction as to where to spend money would be the OPT (which the OP already did) and that plate choke. The MQ-010 would be a fun substitution, and the adjustable bias trim will let you mess with different DCR plate chokes without penalty.
This, of course, goes against the original intent of the schematic, which was almost certainly to provide a relatively easy design to build with parts that aren't super expensive or hard to find.
The chokes I am using are 7.5H and 110 ohms /160ma per channel. Somewhat similar though to the BB-proof, just a little more induction and a little less resistance then the specified Hammond which has a lot more availability.
The drawing takes a little getting used too, but if you look at the other side of the schematic you will see is effectively reservoir capacitors.
That "second" capacitor is split, the top half being the "ultrapath" that terminates at the cathode of the 2A3 above the resistor, as opposed to being terminated to ground below cathode resistor as in more traditional setups. (Like the 6J5)
When John Broskie discussed this circuit he suggested sending
only part of the ultrapath capacitance to the PS, around 20%
and the rest to ground.
It would be easy to try.
Ca
Thanks, I was literally going to look for that page tonight!
I'm finding much smaller value chokes in LCLC arrangement gets good results and is quiet enough , I think this is the best sounding HT supply for SE . I would not suggest the low-value chokes and caps that Jeff does but I would suggest steering well clear of cap-input supplies , even if ultrapath is used . I don't like ultrapath , it always seems more noisy and requires more reservoir capacitance for a given level of ripple which seems to cancel any benefits . YMMV . Why not just stack the supplies ? One for the 2A3 with a slow warmup rectifier and another for the driver stage with SS or DH rectifier ?
Al
Al Noaks
"I'm finding much smaller value chokes in LCLC arrangement gets good results and is quiet enough , I think this is the best sounding HT supply for SE . I would not suggest the low-value chokes and caps that Jeff does but I would suggest steering well clear of cap-input supplies"
I am pretty limited to the CLC configuration I have now, I considered the choke input, but it drops the voltage to much with what I have.
"I don't like ultrapath , it always seems more noisy and requires more reservoir capacitance for a given level of ripple which seems to cancel any benefits . YMMV ."
It's a bit early for me to form a concrete opinion one way or the other with the ultrapath just yet. I don't seem to have an issue as far as noise but I do agree I could use a little more C.
"Why not just stack the supplies ? One for the 2A3 with a slow warmup rectifier and another for the driver stage with SS or DH rectifier ?"
I had not considered that. Just to clarify, did you mean using one transformer or two?
Stacked supplies require 2 transformers or one transformer with two sets of HT windings . Makes for a much more flexible driver stage plus there is a marked reduction in heat and efficiency as there is no large value cathode resistor throwing heat into the air . By the way : resistors are evil
Al
Al,
I once did ( about in 2009 ) a two stage AVVT AV32B DC amp with two separate high voltage windings on one PT ( 9 Ohms DCR ) doing each stage, and it sounded better, more coherent and dynamic with just ONE transformer winding doing both stages, which I postulate MAY have to do with the amp being directly coupled.
That I recall - very well. YMMV, it was just MY one experience, but it cured me of separate supplies on DC two stage amps !!! TOTALLY opposite of my non-DC amp experiences !!
Jeff Medwin
Different from my experience but who am I to question yours ? I would suggest the original poster try both methods and decide for themselves before committing to a chassis . This amp appears to be a 'monkey' choke loaded design . In a stacked setup , a resistively loaded driver will pass the signal through both supplies rather than a choke-loaded design where it will pass through one . I find stacked designs attractive especially WRT flexibility and excessive cathode resistor heat . I especially dislike components doing nothing apart from throw waste heat into the air , the only part that should do that is the HT bleeder
:)
Al
"I would suggest the original poster try both methods and decide for themselves before committing to a chassis . "
You raise some valid points Al Noaks.
As you can see though, test flights have already commenced!
Cheers.
Very nice ! I think Rage is right about your pans :) Are these making noises ?
Al
Thanks guys.
Yes they are making all kinds of noises!
Beautiful amps! Best cake pans I've seen yet.
Rage had in another thread, what would happen to a B+ voltage over a certain amount of time. Based on the in those sims in that discussion.I did a sim that shows what happens in that context, but with this power supply.
This is a 30Hz sine showing a 10 second interval.
Once the B+ recovers from the attack, the voltage stabilizes fine.
Edits: 01/21/15 01/21/15
looks quite a bit different than the other supply.
the other one was essentially ringing after the transient, right?
I thought you interested in seeing the long term stability of the B+ in a stiffer power supply that has more capacitance.
"gargoyle -what happens if you extend out beyond 4 seconds and allow the 150uf cap to discharge?"
I may have misunderstood you.
no we're on the same page... sorry a little under the weather.
I had debated changing driver tubes, parallel 6SN7, different
configurations etc. Rather then speculate, I put the scope on the speaker
terminals to see what kind of voltage I actually need.
95% of my listening is well under 1W, more like 240-500mW. This tells me I
can get away with chopping of some power in the name of quality.At first I debated whether this would be a waste of a 2A3, but in reality
it's not as even if I could push 3.5 watts, I would never need it. 45s may
be an option, but I don't feel compelled to switch.The BB-Proof was running my 2A3s a little hot, due to my higher voltage as
well. and lower OPT voltage drop.So I had to rework the voltages to fall inline with my iron. I am now
running the 2A3s at ~52.5mA/275V. It's a juggle, lowering the current
raises the B+ etc. so this is where I settled.
The 25watt 2A3 cathode resitor is now 4Kohm instead of 3Kohm.Next the driver stage, I was tempted to abandon the BB-proof concept
(Applying negative bias to the 2A3 during start up.), just go a more
traditional route and connect the 6J5 to the regular B+, as it would have
let me use a bigger plate resistor for more swing.
However I persevered and after 20 or so versions later, I settled on a
configuration that works.
Basically it has similar operating points as the BB-Proof, 6mA/~130V on the
plate, but now uses an 8Kohm plate load and a higher B+ of 180V instead of
157V. As Paul Joppa said, this is a good operating point. It works with the
voltages of the amp at least.
I did throw around a few different choke values, but nothing jumps out as
even being close to acceptable, performance, they are garbage lol.The output transformers are now 3500K instead of 2500K.
B+ is now 465VBias on the 6J5 is now stabilized using a 1500uF/50V cathode bypass
capacitor. FWIW I do not consider this to be in directly in the signal
path.The 390 ohm partial bias resistor is now 100ohm to allow for the larger
plate resistor.
With the 6J5 removed in the BB-Proof, the 2A3 saw 191V @ 74mA, -28 bias
resulting in 14watts dissipation.
With the 100ohm, I have 110V @79mA, -8 bias 8.7 watts dissipation.
While I have a touch more current under this fault mode the voltage is less
resulting in a lower wattage, so I should be BB-Proof.The operating points were chosen to get the lowest distortion. Both tubes
have around 2.5% each of mostly 2nd harmonic at full output. the 2A3 bias
is around -52V.
The amps put out 1watt to the speaker terminals.These amplifiers are now stunners. They are now the upgrade I was truly
wanting. They go lower and have better bass then my RH84 (with smallish
OPTs had) They sound "right" as the phase is much much better, this made a
huge difference in the bass department. Transients are much better not
having that delayed bass.I have a square wave of the BB-proof as well, but I will post that after I
square wave test these amps for comparison.I will clean up my schematic and post it here shortly.
Audio Illuminati
Edits: 07/28/15 07/28/15
I'd put another cap after the choke and downsize that ultrapath cap.
Something like 100-300uf after the choke and then the ultra path cap I'd think 40uf would be acceptable.
I'd also eliminate that cap from the cathode to ground.. The 100uf 250v cap.
If you don't like your next round of power supply tweaks maybe you could try one of Jeffs latest creations and give us a report after you've heard it. :)
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: