To use a standard off the shelf OPT with a 32ohm headphone load, can I simply use a transformer rated at 1/2 the primary impedence I want into 8 ohms and then parallel a high wattage 32ohm resistor with the headphones. That is, if I want a 5K primary, I would find a transformer with 2.5K primary and 8 ohm secondary and then parallel a 32ohm resistor with the headphones. Will this damage the frequency response, or performance, etc?
Other than looking at the load lines, is there some trick to picking a primary impedence? Is it just through trial and error that people use a particular primary impedences or is there a specific reason (i.e., for a 45, does 5.3K sound different than 5k -- it is a subjective choice, or does it just maximize power -- an objective choice?)
Are there characteristics of a tube other than the load lines that influence the choice of primary impedence -- specifically, does the plate resistance matter, or something like that?
-d
Other than looking at the load lines, is there some trick to picking a primary impedence? Is it just through trial and error that people use a particular primary impedences or is there a specific reason (i.e., for a 45, does 5.3K sound different than 5k -- it is a subjective choice, or does it just maximize power -- an objective choice?)
Are there characteristics of a tube other than the load lines that influence the choice of primary impedence -- specifically, does the plate resistance matter, or something like that?
-d
Hi dsavitsk. Headphones don't really need much power and most new models appear to be very efficient (I'm in the market and researching a purchase.) I type this while listening to a prototype HP amp and have five diffferent sets of headphones available as loads. The DC impedances vary from ~29 ohm for a Beyers DT 211 TV to ~600 for a pair of vintage AKG 240 Studio Monitors. All but the latter, which are notoriously inefficient, will play loud enough to damage my hearing long term driven by a single 6C45-pe into a 7000 ohm : 6 ohm SE transformer. Checking the manufacturer's specs most call for 100 mW to 50 mW max. They like to see voltage and draw relatively little current.
Measured distortion and frequency response both degrade significantly into the lower impedance phones (very poor OPTs for prototyping) so my recommendation is to leave the secondary unloaded. It's also my general experience that to a point a transformer's performance - measured in any case - improves with reduced drive impedance and reduced load on the secondary. I also found with the current OPT's that a low primary inductance puts a strain on the power supply when reproducing low frequencies. Distortion skyrockets. My recommendation is use the highest impedance primary which gives the required voltage swing. My intention is to go a small James 5K OPT with 8 ohm taps and leave it at that. The high reflected impedance should make the tube happier as well. That said I haven't compared it with other OPTs and would also appreciate any input.
BTW, is that the immortal Dick the Bruiser?
Measured distortion and frequency response both degrade significantly into the lower impedance phones (very poor OPTs for prototyping) so my recommendation is to leave the secondary unloaded. It's also my general experience that to a point a transformer's performance - measured in any case - improves with reduced drive impedance and reduced load on the secondary. I also found with the current OPT's that a low primary inductance puts a strain on the power supply when reproducing low frequencies. Distortion skyrockets. My recommendation is use the highest impedance primary which gives the required voltage swing. My intention is to go a small James 5K OPT with 8 ohm taps and leave it at that. The high reflected impedance should make the tube happier as well. That said I haven't compared it with other OPTs and would also appreciate any input.
BTW, is that the immortal Dick the Bruiser?
It is Reggie "The Crusher" Lisowski (Dick the Bruiser's tag team partner) who sadly died late last year. (http://www.wrestlingmuseum.com/pages/bios/halloffame/crushbio.html) I saw his obit in the LA Times and thought he was an interesting guy, and that it was a great picture. So, hopefully the avatar comes off as respectful as it is meant to be so.
Anyhow, back to the headphones. I am using grado RS-1's which have a 32ohm impedence and require a fair amount of current for phones. I should have stated that earlier as the 32ohm resistor in parallel would have made more sense -- i.e., it would present a 16ohm load which would reflect back and double the input impedence in the example (which actually, this is part of the question -- is this really what happens?)
One of the things I have found with these phones is that they sound way better when the amp can give much much higher current than they really need. For example, using solid state buffers, even though one buf634 is spec'd to give way more current than these could ever use, they sound better with 4 buffers in parallel.
That said, getting every bit of power out of a tube is probably not the most important thing. I am actually planing to use a power triode with no driver tube as I don't need much if any gain, but I do want current capability. For tubes, I am planning to start with 45's and 2a3's, but maybe try a few others along the way including some smaller things like 6j5's or 2c22's. Thus the real question is how important the angle of the load line is other than making sure it fits and allows adequate swing.
Oh, I am also considering some of those james transformers -- I'm curious what you think as compared to other brands?
-d
Anyhow, back to the headphones. I am using grado RS-1's which have a 32ohm impedence and require a fair amount of current for phones. I should have stated that earlier as the 32ohm resistor in parallel would have made more sense -- i.e., it would present a 16ohm load which would reflect back and double the input impedence in the example (which actually, this is part of the question -- is this really what happens?)
One of the things I have found with these phones is that they sound way better when the amp can give much much higher current than they really need. For example, using solid state buffers, even though one buf634 is spec'd to give way more current than these could ever use, they sound better with 4 buffers in parallel.
That said, getting every bit of power out of a tube is probably not the most important thing. I am actually planing to use a power triode with no driver tube as I don't need much if any gain, but I do want current capability. For tubes, I am planning to start with 45's and 2a3's, but maybe try a few others along the way including some smaller things like 6j5's or 2c22's. Thus the real question is how important the angle of the load line is other than making sure it fits and allows adequate swing.
Oh, I am also considering some of those james transformers -- I'm curious what you think as compared to other brands?
-d
Doesn't come across disrespectful at all, great pic. I must admit to some disappointment reading he and The Body were arch rivals. I mean, come on. Jesse!
I'm still not clear the advantage of loading the secondary more than necessary with an additional resistor. I don't have the tools handy to measure impedance vs. frequency, however it's my understanding headphones are a rather benign and constant impedance with frequency, far more so than any speaker. In the case of a typically biased triode reducing the load on the secondary reduces the load on the tube and distortion goes down. Flatter load line and more linear operation. It's easy enough to retrofit a resistor afterwards, but maybe I'm missing something.
Re: current and power I don't think you have anything to worry about with the Grados. It can't possibly be right but every model on the Grado site specs an efficiency in the mid-nineties for an input of 1 mV! Into the rated 32 ohms that calculates as nanowatts. As a sanity check the AKG 701 is 62 ohms, takes a maximum input power of 1/5 watt and puts out 105 dB/volt. One volt into 62 ohms is 16 milliwatts.
Maybe the Grados like a low drive impedance? In that case you're probably into two stages and a lot of NFB to get the output impedance way down. Maybe the BUF isn't crazy about such low loads? Otherwise there's simply no way a 45 and matched OPT won't ring your ears like the Gong Show. If the efficiency specs are close to correct the bigger challenge will be keeping the layout and power supply quiet enough. BTW, are you sure a 45 and OPT pairing have sufficient voltage gain grid to secondary winding? A quick sim in SE AmpCAD shows 10 Vrms in for 1 Vrms out at the secondary.
The James 6123 has only been compared to a defective product, the Hammond old 1628, so I don't have the experience to generalize. It was far and away better than the Hammond and measures very, very well. And I could find nothing approaching it's quality for the price save for maybe a One Electron. I'm more than happy to buy more James.
I'm still not clear the advantage of loading the secondary more than necessary with an additional resistor. I don't have the tools handy to measure impedance vs. frequency, however it's my understanding headphones are a rather benign and constant impedance with frequency, far more so than any speaker. In the case of a typically biased triode reducing the load on the secondary reduces the load on the tube and distortion goes down. Flatter load line and more linear operation. It's easy enough to retrofit a resistor afterwards, but maybe I'm missing something.
Re: current and power I don't think you have anything to worry about with the Grados. It can't possibly be right but every model on the Grado site specs an efficiency in the mid-nineties for an input of 1 mV! Into the rated 32 ohms that calculates as nanowatts. As a sanity check the AKG 701 is 62 ohms, takes a maximum input power of 1/5 watt and puts out 105 dB/volt. One volt into 62 ohms is 16 milliwatts.
Maybe the Grados like a low drive impedance? In that case you're probably into two stages and a lot of NFB to get the output impedance way down. Maybe the BUF isn't crazy about such low loads? Otherwise there's simply no way a 45 and matched OPT won't ring your ears like the Gong Show. If the efficiency specs are close to correct the bigger challenge will be keeping the layout and power supply quiet enough. BTW, are you sure a 45 and OPT pairing have sufficient voltage gain grid to secondary winding? A quick sim in SE AmpCAD shows 10 Vrms in for 1 Vrms out at the secondary.
The James 6123 has only been compared to a defective product, the Hammond old 1628, so I don't have the experience to generalize. It was far and away better than the Hammond and measures very, very well. And I could find nothing approaching it's quality for the price save for maybe a One Electron. I'm more than happy to buy more James.
rdf said:
No, I'm not sure about any of this. That's why asking
So, let me see if I understand what you are saying. Not only does the OPT step down impedence, but also voltage (in a ratio of ~ 10:1) which means that the tube before the transformer needs to output 10x higher V than I want at the output?
My source puts out ~ 2Vrms. The 45 has a gain of about 3.5 which results in an output of about .7V. For normal people, this is probably way too low. However, I never turn up my headphones loud at all (never past about 8:30 on my melos -- really, i turn it to the lowest setting at which the volume pot tracks evenly) so it may not be critical. Iin fact, even with just a buffer I rarely turn the volume past about 10:00. So, assuming that the V is high enough for me, so long as the tube can supply sufficient current at that voltage I am okay? Are there other things to consider?
As for the PS, it is worse than you think. I am planning to use an ultrapath design which is notorious for amplifying hum on the heater. For B+, I am planning on a RCLCRC filter. For each filament, a CLC with about 80000uF in total. In the Duncan labs software this gets the ripple well into the uV range for both.
As I say, though, this is all new to me, so if there are elements that I need to consider then I need to hear them before I start buying stuff as this project is likely to not be cheap. It may be that the 45 just won't work here, but if it will, I'd like to give it a shot just to do something a little different.
I attached a schematic to show what I am thinking. The R in the filament filter should be a choke, however.
One other thing, the main reason I am insistant about putting a resistor in parallel with the headphones is so that if it gets turned on without phones plugged in it won't cause a problem. Maybe this is less of an issue that I think?
Attachments
![ultrapath45[2].gif](/community/data/attachments/36/36269-862fcef31eec886098b14c5fe6737a66.jpg)
I have Grado 32R phones too. I have not ruled out a 32R in parallel and then running on the 16R primary. It side-steps the whole ratio-ing thing, and uses the whole secondary too.
I do like overkill, so a well driven pair of 7233 may just be my headphone final of choice. Plenty of light 5-7k a-a outputs with 16R secondaries.
The other option is an OPTx with 62.5R secondaries, and put a 32R in series. I have one of these in 3, 5, 6k6, 8 and 10k a-a in the Peerless 20-20 and UTC LS flavours. Can't decide which is better, everything in parallel for ~60R, or everything in series for 16R.
There is nothing a small surplus of power can't take care of...
cheers,
Douglas
I do like overkill, so a well driven pair of 7233 may just be my headphone final of choice. Plenty of light 5-7k a-a outputs with 16R secondaries.
The other option is an OPTx with 62.5R secondaries, and put a 32R in series. I have one of these in 3, 5, 6k6, 8 and 10k a-a in the Peerless 20-20 and UTC LS flavours. Can't decide which is better, everything in parallel for ~60R, or everything in series for 16R.
There is nothing a small surplus of power can't take care of...
cheers,
Douglas
2.5V 1.5amp filament on the 45. You might consider a bridge rectified LCRC or LCLC instead of a full wave CLC. Critical inductance for one 45 is only 1.67mH.
If the 2 45 heaters are paralleled , critical inductance is .84mH, giving an "optimal" filter with 1.7mH and allowing for a heater to open.
It will be electrically quieter and less strain on the filament transformer than a full wave CLC. It also has less diode losses, which are significant at these voltages. It drops more of the excess voltage in filters instead. Use Schottkey diodes here, they are cheap at these powers.
If the 2 45 heaters are paralleled , critical inductance is .84mH, giving an "optimal" filter with 1.7mH and allowing for a heater to open.
It will be electrically quieter and less strain on the filament transformer than a full wave CLC. It also has less diode losses, which are significant at these voltages. It drops more of the excess voltage in filters instead. Use Schottkey diodes here, they are cheap at these powers.
Tweeker said:
I don't think I know what critical indictance is. Can you explain it more?
Whether the heaters are parallel or not depends upon what I can find for a power transformer . I am starting with 45's, but may try other tubes including 2a3's that will need more current and some others, some of which will need 6.3V. So, there is supposed to be a voltage dropping resistor in there as well.
A criticaly sized inductor is one that for a given frequency, waveform, and current, outputs the average voltage. Or something like that. If the inductor is less than critical, voltage will rise above this value. The amount of inductance required to be critical decreases with greater current. For a sine wave the average voltage is .637*peak voltage and .9* rms voltage.
So if you had a critically sized inductor on a 1,000VAC RMS bridge rectifier, it would output about 900VDC, (.637*1414V peak or .9*1000VAC RMS).
If the current drops below that value which the inductor is critical, or the inductance drops below critical, the voltage will rise. Swinging chokes take advantage of this to give better regualtion by dropping in inductance from dc saturation as current rises such that the output voltage drops little.
Inductors oppose changes in current and in effect drag voltage up and down to do this. Output is a steady current rather than the spikes seen with capacitor input.
You could also change between choke input LCRC/LCLC layouts and CLC to give higher voltage with the same transformer.
So if you had a critically sized inductor on a 1,000VAC RMS bridge rectifier, it would output about 900VDC, (.637*1414V peak or .9*1000VAC RMS).
If the current drops below that value which the inductor is critical, or the inductance drops below critical, the voltage will rise. Swinging chokes take advantage of this to give better regualtion by dropping in inductance from dc saturation as current rises such that the output voltage drops little.
Inductors oppose changes in current and in effect drag voltage up and down to do this. Output is a steady current rather than the spikes seen with capacitor input.
You could also change between choke input LCRC/LCLC layouts and CLC to give higher voltage with the same transformer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.