Hi All,
After finishing up my pair of Mark Audio 10p Pensils i've set about making a low wattage tube amp to drive them. Aside from a handful of guitar my only tube hifi experience is here ->http://www.diyaudio.com/forums/tubes-valves/256704-12bh7-headphone-amp.html
The MA 10p's are not particularly efficient at 89db/W but I also live in a small apartment so I figured something in the range of 5 watts would suffice. After googling around for se amps in that range I decided to start small with EL84 (though triode mode EL34, KT88, etc were options). Mixed reviews exist of SE power pentodes due to low damping factor and higher distortion than triode mode but I had stumbled upon this tubeCAD post that fascinated me:
Tube-based Buffer amplifier
By using the secondary of the OT as degenerative cathode feedback you get a better output Z and lower distortion at the cost of voltage gain - but you keep the max power of pentode/tetrode. So, after playing with the numbers and simulating the third circuit (the "complex" buffer with a driver tube) I decided to give it a shot. The el84 requires very little drive voltage as is so a small increase in the needed signal is easier to mange than with say, the 300b example that needs 75Vp! I figured if trying this was disaster I could just regroup and build something more conventional with the parts.
I'm using an ECC99 as a low output impedance, high-linearity, modest-gain driver stage with a voltage gain of about 14. That's plenty of feedback (~24db?) that should push the Zout to below .25 ohms. Simulation claims
rolleyes <.2%THD at 5W into 6.2 ohms (on the 8 ohm tap). Oh ya, the MA drivers are 6.2 ohms.
...and the prototype:
Not much to look at! Currently I'm driving this directly with the headphone out of an ipod classic and already it gets loud enough for regular home use with louder recordings. It sounds quite good and I bet anyone with a solid preamp could call it finished. I don't have a preamp and I'd prefer one package so the next version will include a single gain stage to increase input sensitivity.
So, that's what's on the docket for soldering this week. I'm open to feedback and hoping to hear other's thoughts. Just wanted to share a small SE project that looked a little different from the conventional small-scale SE designs.
-Brian
After finishing up my pair of Mark Audio 10p Pensils i've set about making a low wattage tube amp to drive them. Aside from a handful of guitar my only tube hifi experience is here ->http://www.diyaudio.com/forums/tubes-valves/256704-12bh7-headphone-amp.html
The MA 10p's are not particularly efficient at 89db/W but I also live in a small apartment so I figured something in the range of 5 watts would suffice. After googling around for se amps in that range I decided to start small with EL84 (though triode mode EL34, KT88, etc were options). Mixed reviews exist of SE power pentodes due to low damping factor and higher distortion than triode mode but I had stumbled upon this tubeCAD post that fascinated me:
Tube-based Buffer amplifier
By using the secondary of the OT as degenerative cathode feedback you get a better output Z and lower distortion at the cost of voltage gain - but you keep the max power of pentode/tetrode. So, after playing with the numbers and simulating the third circuit (the "complex" buffer with a driver tube) I decided to give it a shot. The el84 requires very little drive voltage as is so a small increase in the needed signal is easier to mange than with say, the 300b example that needs 75Vp! I figured if trying this was disaster I could just regroup and build something more conventional with the parts.
I'm using an ECC99 as a low output impedance, high-linearity, modest-gain driver stage with a voltage gain of about 14. That's plenty of feedback (~24db?) that should push the Zout to below .25 ohms. Simulation claims
rolleyes <.2%THD at 5W into 6.2 ohms (on the 8 ohm tap). Oh ya, the MA drivers are 6.2 ohms. ...and the prototype:
An externally hosted image should be here but it was not working when we last tested it.
Not much to look at! Currently I'm driving this directly with the headphone out of an ipod classic and already it gets loud enough for regular home use with louder recordings. It sounds quite good and I bet anyone with a solid preamp could call it finished. I don't have a preamp and I'd prefer one package so the next version will include a single gain stage to increase input sensitivity.
So, that's what's on the docket for soldering this week. I'm open to feedback and hoping to hear other's thoughts. Just wanted to share a small SE project that looked a little different from the conventional small-scale SE designs.
-Brian
Last edited:
interesting ideas !!
...but sadly the photos didn´t show up...
Whoops! Re-hosted. Thanks
Last edited:
Hi Eli,
Ok, I'll take another look at an open loop front end. Using the 12ax7's surplus gain was deliberate as it specc'd the best in sim... but that doesn't mean it would sound the best. I'll see if I can get my hands on a 6922 or equivilent. I am trying to avoid the complexities of ccs stages right now. Wouldn't a ccs loaded 6922 have close to the same gain as a 12ax7? A gain approaching mu (33) for the former and ~33 with latter using the normal 100k/1k unbypassed configuration?
The speaker is in parallel with the secondary whose dc resistance will be far lower, so the dc offset caused by cathode current will be across a fraction of an ohm, not 6-8 ohms. I had intended to measure this though and will report back. I suppose a work around would be to ground the cathode resistor and tie the secondary back through a bypass cap.
Sent from my SM-N900V using Tapatalk
Yes, I didn't consider that . But then it also means some dc flows through the secondary coil. Need to look at this carefully. A capacitor will stop the dc but then using a cap in series with the signal is always a questionable factor !
However it's an interesting idea and maybe you could try it out in bits at first. You could simulate it of course but with transformers it might be worth while building the input stage and add the speaker and transformer . Not sure what 10 mA dc will do to the core. Will look it up !
However it's an interesting idea and maybe you could try it out in bits at first. You could simulate it of course but with transformers it might be worth while building the input stage and add the speaker and transformer . Not sure what 10 mA dc will do to the core. Will look it up !
I guess the risk is saturation of the transformer secondary. I have no idea how to predict the affects, but I do now recall preferring the "complex buffer" to the simple one as the latter puts the whole power tube idle current through the secondary. They're 7 watt transformers and I'm using the 8 ohm tap so I'm using 1% of the minimum peak current they are rated for.Yes, I didn't consider that . But then it also means some dc flows through the secondary coil. Need to look at this carefully. A capacitor will stop the dc but then using a cap in series with the signal is always a questionable factor !
However it's an interesting idea and maybe you could try it out in bits at first. You could simulate it of course but with transformers it might be worth while building the input stage and add the speaker and transformer . Not sure what 10 mA dc will do to the core. Will look it up !
Sent from my SM-N900V using Tapatalk
BJosephs,
You are correct about a CCS loaded 6922 section yielding a gain of 33, before the grid to ground resistance of the next stage is taken into account. That resistance becomes, for all practical purposes, the NET AC load the 6922 section "sees". Size that part at 2X the recalculated (necessitated by the unbypassed cathode bias resistor and the associated local current NFB) 6922 RP and 6922 section stage gain will be approx. 16X.
You are correct about a CCS loaded 6922 section yielding a gain of 33, before the grid to ground resistance of the next stage is taken into account. That resistance becomes, for all practical purposes, the NET AC load the 6922 section "sees". Size that part at 2X the recalculated (necessitated by the unbypassed cathode bias resistor and the associated local current NFB) 6922 RP and 6922 section stage gain will be approx. 16X.
Oh, ok. I didn't know the grid resistor of the next stage was being reduced too.
Sent from my SM-N900V using Tapatalk
Sent from my SM-N900V using Tapatalk
Well, I built up the second version using the 12AX7 front end and I did not like what I heard. The sensitivity went up but something went all wonky with the bass - not sure why. The first version is great, I may box it up as is and live with the lower sensitivity until I make a preamp... or go with another option for the front end like the 6922 Eli suggested.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.