Just starting to design my first SE amp probably based on RH84 rev2
I'm using Duncan's PSU Designer II to model the power supply and LTSpice for the amp.
At the moment my power supply design (2-pole RC) is showing approximately 0.1V of 100Hz ripple when on load (100mA).
Is this a reasonable level to aim for or do I need to do better (more expensive)
Thanks
I'm using Duncan's PSU Designer II to model the power supply and LTSpice for the amp.
At the moment my power supply design (2-pole RC) is showing approximately 0.1V of 100Hz ripple when on load (100mA).
Is this a reasonable level to aim for or do I need to do better (more expensive)
Thanks
Last edited:
Split one of your existing RC stages in half, making it into two RC stages (each "R" half the resistance of the original, same value capacitors though is fine) for some additional filtering, and give it a listen.
With an SE output stage, and on efficient speakers 0.1V of ripple may be audible at near field, or in a very quiet listening space.
With an SE output stage, and on efficient speakers 0.1V of ripple may be audible at near field, or in a very quiet listening space.
Thanks Nigel
There is another 10K into 100uF in the design going to the 22K anode resistor of the 12AT7 which according to the model reduces ripple to that to less than 1mV so that bit looks OK.
I'm intending to use a separate 6.3V transformer and semiconductor bridge rectified 1:1 isolating transformer. That will allow me to use a PIC powered by the 6.3V supply to switch the first HT resistor into circuit once a high value bleed resistor has very slowly brought up the HT rail. This should give a nice soft start for the valves and avoid any high power across the first HT resistor.
So soft start is 10K into 470uF followed by 68R into 470uF. Once the PIC senses the HT rail is above 290V it will then switch a 22R resistor in place of the 10K one.
If my calcs are correct this will give an HT rail of 104mA at 312VDC with ripple of 0.06V
There is another 10K into 100uF in the design going to the 22K anode resistor of the 12AT7 which according to the model reduces ripple to that to less than 1mV so that bit looks OK.
I'm intending to use a separate 6.3V transformer and semiconductor bridge rectified 1:1 isolating transformer. That will allow me to use a PIC powered by the 6.3V supply to switch the first HT resistor into circuit once a high value bleed resistor has very slowly brought up the HT rail. This should give a nice soft start for the valves and avoid any high power across the first HT resistor.
So soft start is 10K into 470uF followed by 68R into 470uF. Once the PIC senses the HT rail is above 290V it will then switch a 22R resistor in place of the 10K one.
If my calcs are correct this will give an HT rail of 104mA at 312VDC with ripple of 0.06V
Add another resistor/cap, or grab a $15 choke if it truly bothers you. Or build a Mosfet ripple filter (cheap, easy, very effective, $5USD will make one that is dead silent and that will be much more cost effective than any choke of four times the cost) and be done with it. A crappy PSU makes a good amplifier circuit sound crappy.
I run this in my 6sn7 push pull, change that preceding resistor to a smaller value, and may be increase the 12k resistance to a 18~22k for more headroom. Easy build, parts are cheap. Dead silent, even with 97dB/w speakers. Much more cost effective than a choke, and no stray magnetic fields to fool with. Nice little soft start feature too.
This is indeed cheaper than using more caps, even if they are cheap Chinese ones. How much of a heatsink does that MOSFET need?
On my particular use, its just a few degrees above ambient with a small 1"x1"x2" (one of these, actually) or so sink, but I'm yanking maybe 50mA total through it, with maybe a 10~12 volt drop across the Fet. For higher current use I would want to drop maybe 10% of the voltage supplied through it for safe margins, but less would work well on a class A design like the circuit the OP is using.
I've been meaning to bolt it straight to my chassis, but need to strip some paint so I can mount it. I think page six of my 6SN7 thread is where GoatGuy and I were riffing back and forth on the basics for it.
I've been meaning to bolt it straight to my chassis, but need to strip some paint so I can mount it. I think page six of my 6SN7 thread is where GoatGuy and I were riffing back and forth on the basics for it.
at a plate voltage of say 300 volts, will 0.1v ripple matter that much? i hardly think it will...
I had to simulate and build that regulator to convince myself it would work but I'll be a monkey's uncle if a handful of parts worth less than a couple bucks does in fact do an excellent job. So much for my usual standby of an LM317 zener protected regulator with a high voltage parallel pass transistor.
It is cheaper to do some thinking and figuring before you open your wallet.
What is the power supply rejection ratio (PSRR) of a "SE amp"? It depends; Pentodes are much better than triodes. But I will assume triode.
We usually ballpark the tube's plate resistance (rp) and then use a 2X-to-5X rule of thumb to get a load impedance. 2X is more power, 5X is less THD, though the 2X amp may play soft as clean as the 5X amp. Let's take 3X.
So for tube of rp=2K, we use 6K load.
The ripple splits as in any resistor voltage divider. 1/4 across the 2K, 3/4 across the 6K.
We will hear the ripple in the 6K. So call it 75mV. (Peak? RMS? Peak-to-peak??)
That is divided-down by the transformer ratio. 6K:8 is 750:1 impedance, 27:1 voltage. Call it 2.8mV at the 8 Ohm side.
This is (too conveniently!) 1/1000 of the 2.8V (1 Watt) we use for rating speakers. Say your speakers are 90dB SPL at 1W at 100Hz in a room-size space.
1/1000 of voltage is 60dB. So that makes 30dB SPL of ripple in the room.
Is this low enough? I've heard worse. Due to Fletcher-Munson, 30dB SPL at 100Hz is not real audible. I would worry more about your higher harmonics. However you say 2-stage filter, and that will *tend* to cut the higher harmonics quickly (more info needed).
Personally I would plan (leave space for) another 20dB (10:1) drop of ripple, because I know I would note 30dB SPL at 120Hz when I put my head in the speaker. I don't demand no-buzz with my ear on the grille, but it is nice.
I've got one included on my PCB, I'm so used to using them that I don't draw them anymore. They are just "assumed" to be there.
Anything from 12-15 volts will usually work.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.