Hi All,
So, I've done about a dozen tube instrument amp projects and I'm thinking it's time to branch out and try something in the hifi type world. Having very little experience trying minimize tube distortion I figured I'd start small and make myself a little tube headphone amp... maybe even spring for a new set of headphones too once I get done with it.
I like transformers, and I don't mind paying for a decent pair, so I figured I'd do a small SE design using an off-the-shelf OT - Sowter and Lundhal are the two makers I've looked at, the latter of which is pushing it a bit budget wise. I've been googling the subject of transformer-coupled SE headphone amps and I'm surprised how little is out there. Maybe I'm not looking hard enough or I'm searching the wrong terms... everything seems to be OTL designs.
So I thought I'd finally learn to plot a load line and try to put a plan together myself. I'm afraid I'm a bit uninitiated outside of the 12*** world so I picked the 12BH7 as a starting point since it can handle some current and it seems liked, at least in voltage gain stages. And I can get them reliably.
So, I plotted a load line (Basically my first one!) based on the "typical" (datasheet) characteristics of 11.5mA@250V - it worked out to a 22k load impedance and about .5W of output power:
I went by this http://www.blueguitar.org/new/text/threads/from_mark/SBench_Loadlines_Series.pdf instruction. Sorry for using pencil...
I'm thinking I should have a bit more headroom so I figured I'd parallel both halves of the tube and use 1/2 the impedance to get about 1W output. Sowter sells the 8988 OT which is 10k:600 (And down) and is good for 2W @ 50hz. Seems like a good match and a reasonable starting point.
And it looks like an incoming swing of about 20Vp-p at the grid yields about 280Vp-p at the anode, voltage gain of 14.... I'll need a driver stage to get to that 20V though. Looking for suggestions on that part...
So am I way off or off to a decent start?
Regards,
Brian
So, I've done about a dozen tube instrument amp projects and I'm thinking it's time to branch out and try something in the hifi type world. Having very little experience trying minimize tube distortion I figured I'd start small and make myself a little tube headphone amp... maybe even spring for a new set of headphones too once I get done with it.
I like transformers, and I don't mind paying for a decent pair, so I figured I'd do a small SE design using an off-the-shelf OT - Sowter and Lundhal are the two makers I've looked at, the latter of which is pushing it a bit budget wise. I've been googling the subject of transformer-coupled SE headphone amps and I'm surprised how little is out there. Maybe I'm not looking hard enough or I'm searching the wrong terms... everything seems to be OTL designs.
So I thought I'd finally learn to plot a load line and try to put a plan together myself. I'm afraid I'm a bit uninitiated outside of the 12*** world so I picked the 12BH7 as a starting point since it can handle some current and it seems liked, at least in voltage gain stages. And I can get them reliably.
So, I plotted a load line (Basically my first one!) based on the "typical" (datasheet) characteristics of 11.5mA@250V - it worked out to a 22k load impedance and about .5W of output power:
An externally hosted image should be here but it was not working when we last tested it.
[/URL]
I went by this http://www.blueguitar.org/new/text/threads/from_mark/SBench_Loadlines_Series.pdf instruction. Sorry for using pencil...
I'm thinking I should have a bit more headroom so I figured I'd parallel both halves of the tube and use 1/2 the impedance to get about 1W output. Sowter sells the 8988 OT which is 10k:600 (And down) and is good for 2W @ 50hz. Seems like a good match and a reasonable starting point.
And it looks like an incoming swing of about 20Vp-p at the grid yields about 280Vp-p at the anode, voltage gain of 14.... I'll need a driver stage to get to that 20V though. Looking for suggestions on that part...
So am I way off or off to a decent start?
Regards,
Brian
Check out my headphones amp design here:
http://www.diyaudio.com/forums/tubes-valves/191632-improved-tube-headphones-amp.html
Cheers
Ian
http://www.diyaudio.com/forums/tubes-valves/191632-improved-tube-headphones-amp.html
Cheers
Ian
Ian,
You know what? I just might. Not sure why I'm so in love with the idea of an SE design, but without a distortion analyzer I'm probably not well equipped to really break new ground. A proven design would be a smart choice and the transformers are even a bit cheaper.
Does the SRPP stage require an trimming or would I get good performance just dropping in the same values you used? Do the top and bottom halves of the EC99 require matching? I'm thinking not, but the pairs for both channels probably do (including the 12AX7).
Do you have any feedback on the plots I posted? I'd still like to figure out if I'm grasping the load line concept.
Brian
You know what? I just might. Not sure why I'm so in love with the idea of an SE design, but without a distortion analyzer I'm probably not well equipped to really break new ground. A proven design would be a smart choice and the transformers are even a bit cheaper.
Does the SRPP stage require an trimming or would I get good performance just dropping in the same values you used? Do the top and bottom halves of the EC99 require matching? I'm thinking not, but the pairs for both channels probably do (including the 12AX7).
Do you have any feedback on the plots I posted? I'd still like to figure out if I'm grasping the load line concept.
Brian
Hi Brian,
All topologies are a compromise of some sort. The SE has its own problems, in my view the biggest being the need to have dc flowing in the output transformer. This means it must be gapped which lowers its inductance reulting is poorer bass response. Fine for guitar amps but not for headphones. of course you can get one with a decent inductance but it will be very big and expensive.
No special trimming is required for the SRPP. it is a meaty stage that will have a good go at driving almost anything. Its drawback is its relatively high distortion which is reduced 10 fold in my design by the negative feedback. For the same reason there is no need to do any tube matching. The gain is determined by passive components not the tubes themselves.
For an SE design, the dc loadline is almost vertical because the dc resistance of the transformer is very low. So you can start by drawing a vertical line at the chosen HT voltage. Your operating point will be somewhere on this line. You then need to draw an ac load line with a slope equal to the load reflected back to the primary of the output transformer. You ten side this ac load line up and down untill you find a suitabgle intersection with the dc load line that llies within the dissipation of the tube.
The Valve Wizard explains it much better than I can:
The Valve Wizard -Single Ended
Cheers
Ian
All topologies are a compromise of some sort. The SE has its own problems, in my view the biggest being the need to have dc flowing in the output transformer. This means it must be gapped which lowers its inductance reulting is poorer bass response. Fine for guitar amps but not for headphones. of course you can get one with a decent inductance but it will be very big and expensive.
No special trimming is required for the SRPP. it is a meaty stage that will have a good go at driving almost anything. Its drawback is its relatively high distortion which is reduced 10 fold in my design by the negative feedback. For the same reason there is no need to do any tube matching. The gain is determined by passive components not the tubes themselves.
For an SE design, the dc loadline is almost vertical because the dc resistance of the transformer is very low. So you can start by drawing a vertical line at the chosen HT voltage. Your operating point will be somewhere on this line. You then need to draw an ac load line with a slope equal to the load reflected back to the primary of the output transformer. You ten side this ac load line up and down untill you find a suitabgle intersection with the dc load line that llies within the dissipation of the tube.
The Valve Wizard explains it much better than I can:
The Valve Wizard -Single Ended
Cheers
Ian
Me too. I think you are searching in the wrong places.
ecp audio diy
TORPEDO Overview
Most headphones are very efficient and don't need much power. A little headroom is not a bad thing, but no reason to shoot for a full watt. Sennheiser HD800s, for instance, are 102dB per Vrms. That works out to ~3mW.
12BH7 sections in || with a 10K load seems like a pretty decent option to me. And, finding transformers with a 22K impedance that work well will be a challenge.
Thanks for the links, dsavitsk, I'll check them out.
The mention of ECC99 prompted me to repeat my load line exercise this morning. I'm a bit confused about how the numbers work out… Should I be plotting load lines for the transformer load (let's say 10k in this case) and one tube, then expecting double power w/ two tubes into the same load; or plotting the load line with 2xLoad when I know I would intend to parallel two sections?
Brian
The mention of ECC99 prompted me to repeat my load line exercise this morning. I'm a bit confused about how the numbers work out… Should I be plotting load lines for the transformer load (let's say 10k in this case) and one tube, then expecting double power w/ two tubes into the same load; or plotting the load line with 2xLoad when I know I would intend to parallel two sections?
Brian
...And on the subject of an AC coupled output driven by an SRPP, both Jensen and Cinemag (transformer companies I've used before for DIY studio stuff) have 10k:600 output transformers. The Jensen is $140 (About as much as the Sowters) but the Cinemag looks to be less than $100:
http://cinemag.biz/output/PDF/CM-9589.pdf
And it looks to be able to handle +23dbu @ 30hz into 600 ohms with a 2.4k driving source and ~.005%THD… If my math is right that works out to about 11V and 200mW? Maybe an option? Distortion is better with lower source impedances and the ECC99 SRPP works out to about 1400 ohms (before accounting for the feedback). So, maybe a decent choice for a local option?
I should have mentioned that my best set of headphone are beyerdynamic DT 700 600ohm, so it's not a bad call to plan around 600Ohm only right now.
Brian
http://cinemag.biz/output/PDF/CM-9589.pdf
And it looks to be able to handle +23dbu @ 30hz into 600 ohms with a 2.4k driving source and ~.005%THD… If my math is right that works out to about 11V and 200mW? Maybe an option? Distortion is better with lower source impedances and the ECC99 SRPP works out to about 1400 ohms (before accounting for the feedback). So, maybe a decent choice for a local option?
I should have mentioned that my best set of headphone are beyerdynamic DT 700 600ohm, so it's not a bad call to plan around 600Ohm only right now.
Brian
Last edited:
Math looks right. If you are driving 600R phones, then there are lots of good options. Cinemags are great. And they'll butt stack them so you can use them single feed -- though I don't see a spec for how much DC is OK.
If you are thinking parafeed, however, you can get cheapo Edcors to experiment with, and then buy good stuff later.
Do keep in mind that there aren't really [m]any great 600R phones out there, so you might want to plan for more future options.
If you are thinking parafeed, however, you can get cheapo Edcors to experiment with, and then buy good stuff later.
Do keep in mind that there aren't really [m]any great 600R phones out there, so you might want to plan for more future options.
If you want excellent quality 10K:600 output transformers at a reasonable price then I would suggest the Carhnill VTB2291 for a capacitively coupled output. I think they also do a gapped version if you want an SE version.
If you are driving 600 ohm phones then you do not really need something as beefy as the ECC99. A regulat 6922 (industril version of the 6DJ8/ECC88) in an SRPP will be fine.
After developing the headphones amp, I applied the same basic principles to a mic pre which uses a 6922 SRPP output stage and a 12AX7 input stage modified to allow the gain to be varied from 6 to 40dB. It is used as the basis of my EZTubeMixer project at groupdiy.com. It used the VTB2291 output transformer wired 2K4:600 and will happily ouptut +20dBm into a 600 ohm load. Schematic is here:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/EurocardCCTsht3.jpeg
Cheers
Ian
If you are driving 600 ohm phones then you do not really need something as beefy as the ECC99. A regulat 6922 (industril version of the 6DJ8/ECC88) in an SRPP will be fine.
After developing the headphones amp, I applied the same basic principles to a mic pre which uses a 6922 SRPP output stage and a 12AX7 input stage modified to allow the gain to be varied from 6 to 40dB. It is used as the basis of my EZTubeMixer project at groupdiy.com. It used the VTB2291 output transformer wired 2K4:600 and will happily ouptut +20dBm into a 600 ohm load. Schematic is here:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/EurocardCCTsht3.jpeg
Cheers
Ian
OK, so there are no tricks in this. Any 10k:600 with enough output and low distortion would all technically 'work'. Thanks.
Good advise. Future proofing is worth consideration… plus I spent a hour yesterday drawing up a switching arrangement for a 3 or 4 winding secondary like Sowter uses.
Ian, I've seen your project page. I have to say you only make these things look "EZ"! I noticed you dropped the grid stoppers from Pete Millet's design, did you find them to be unnecessary? Also it looks like you power the driver and SRPP stages from the same PS node rather than decouple with a resistor and cap. Did you use the same 470uF that PM used?
I'm going to pull the trigger on parts before the weekend is over.
Thanks,
Brian
You definitely do not need a grid stopper in the top triode of the SRPP as it is configured for unity gain. It is possible you might need one in the grid of the bottom triode of the ECC99 to prevent parasitic oscillation but I have not found it necessary. If you are doing a P2P build then it is easy enough to include a 1K grid stopper just to be on the safe side.
I did not separately decouple the 12AX7 stage in the headphones amp design but in the subsequent mic pre design I did decouple it with a 10K resistor and 10uF or 22uF capacitor. This is partly because the 12AX7 stage has a worse PSRR then the SRPP stage so it benefits from extra decoupling and the last thing you want in a mic pre is PSU noise in the first stage. If I were building the phones amp again I would include them this time.
My power supply is different from Pete's. I use 4 x 470uF 450V caps in a CRCRCRC combination to get a really low noise HT supply. I usually select the resistors to have about 10V drop across them at the nominal current draw. In this case it is close to 30mA for two channels which makes them 330R.
Cheers
Ian
OK, thanks. Did you ever consider a series pass regulator to get rid of the ripple? I have some power mosfets and 300v zeners that I think I'll try. My power transformer has a 260V winding @ 100mA, full wave rectified and filtered it will be about 360VDC @ .062mA. I have 100uf+100uf and 50uf+50uf JJ can caps that I'll use during prototyping… I have them to use and I can upgrade later. So it would go FWB>100uf>IRF820 series pass regulator>100uf then two small resistors leading to each of the 50uf sections… one for the left channel and one for the right. If I get PS noise I have some axial caps that I can put put in to decouple the AX7 stages. Thats the plan for now.
I didn't see any notes on how much current you bias the SRPP for, is it about 15mA based on your total of 30mA current draw from an earlier post? Pete millet has 20mA on his schematic next to each SRPP.
Thanks for all the help,
Brian
Last edited:
I would definitely use the regulator over a string of RC or LC filters. However, I would not add another RC filter after the reg. A smallish (or large if you like) cap to ground from the source is all you need.
If you do use an SRPP type topology, at least look into the Beta follower circuit. SRPP is really not a great circuit (compared to a CCS loaded common cathode, for instance), and the B follower corrects at least some of that.
If you do use an SRPP type topology, at least look into the Beta follower circuit. SRPP is really not a great circuit (compared to a CCS loaded common cathode, for instance), and the B follower corrects at least some of that.
I considered them but my personal preference is for as few semiconductors as possible in my designs. I have built a couple of regulators using the TL783 but, even with protection diodes, I found them incredibly easy to blow up. A three stage RC filter can easily outperform any regulator in terms of output ripple. Regulators can also be quite noisy due to their high internal gain.
My SRPP is identical to Pete Millet's. The idle current depends on the HT voltage. In the prototype, my HT was probably a little higher than Pete's. I measured the cathode current by measuring the voltage across the 300 ohm cathode resistor. It was about 4.5V hence the 15mA figure quoted.
Cheers
Ian
Actually a simple triode strapped 4P1L SE output stage is even better. A 4P1 L only costs few dollars right now. A good quality SE output transformer will do the job quite nicely.
It's not the ripple that matters (both can knock it down below inaudible), it's the output impedance.
Ian,
I was reading this (http://www.valvewizard.co.uk/SRPP_Blencowe.pdf) article by the Valve Wizard and I spotted some stuff that I may try when I knock my prototype together:
1. The current in the top and bottom triode are imbalanced without an anode resistor (equal to the 2 cathode resistors) connected between B+ and the top triode. It wasn't clear in the text if this has an impact on distortion.
2. The distortion goes down (as well as gain and PSRR) with the lower cathode resistor unbypassed - obvious it seems but maybe there is sufficient loop gain already that we can throw some away in favor of optimizing the the output stage and dumping 1 capacitor.
3. The cathode resistor values are a function of expected load, and the values in the is circuit are not what Merlin's formulae suggest (though they seem to optimize ECC99 idle current). Under "correct" conditions the transformer provides 10k AC load with works out to 1K1 resistors (Too cold, 5.6mA). Since the Sowter 8665 has a range of impedances for each tap the worst case is an actual 3k6 load (100 ohm phones on the 100 - 250 secondary) the "optimal" cathode resistor is more like 463 ohms (Idle current drops from 17mA to 12mA) - who knows what's better (higher idle current or better matching to the load), but I thought I'd throw that out there.
Brian
I was reading this (http://www.valvewizard.co.uk/SRPP_Blencowe.pdf) article by the Valve Wizard and I spotted some stuff that I may try when I knock my prototype together:
1. The current in the top and bottom triode are imbalanced without an anode resistor (equal to the 2 cathode resistors) connected between B+ and the top triode. It wasn't clear in the text if this has an impact on distortion.
2. The distortion goes down (as well as gain and PSRR) with the lower cathode resistor unbypassed - obvious it seems but maybe there is sufficient loop gain already that we can throw some away in favor of optimizing the the output stage and dumping 1 capacitor.
3. The cathode resistor values are a function of expected load, and the values in the is circuit are not what Merlin's formulae suggest (though they seem to optimize ECC99 idle current). Under "correct" conditions the transformer provides 10k AC load with works out to 1K1 resistors (Too cold, 5.6mA). Since the Sowter 8665 has a range of impedances for each tap the worst case is an actual 3k6 load (100 ohm phones on the 100 - 250 secondary) the "optimal" cathode resistor is more like 463 ohms (Idle current drops from 17mA to 12mA) - who knows what's better (higher idle current or better matching to the load), but I thought I'd throw that out there.
Brian
Last edited:
And here's a switching diagram for selecting impedance range. One channel shown:
The top line connects to the tip or ring, the bottom is common. As shown it's in the 30-100 ohm position with all three coils in parallel. It requires a 8P3T rotary switch for 2 channels. I doubt I'll pursue it until I've got the circuit working well, but I did want to draft it up to see if it was possible with the this transformer secondary arrangement.
Brian
An externally hosted image should be here but it was not working when we last tested it.
[/URL]
The top line connects to the tip or ring, the bottom is common. As shown it's in the 30-100 ohm position with all three coils in parallel. It requires a 8P3T rotary switch for 2 channels. I doubt I'll pursue it until I've got the circuit working well, but I did want to draft it up to see if it was possible with the this transformer secondary arrangement.
Brian
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.