|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
195.93.21.66
In Reply to: Raven,Akido,Ultrapath line amp posted by Rune on April 28, 2007 at 03:39:25:
If you want the best sound, I (and others) would suggest using directly heated triodes. There's a bit of an art to chosing and using them, but they don't have to be a fright for the constructor. You need a good DC supply for each tube (current source best), and to get the best out of them a 100k stepped attenuator on the input and a nice Lundahl (or other) transformer on the output. Virtually nothing in all audio is better than or even as good as this. But even better is to get a differential DAC like the RAKK DAC and put it straight into the grids of a differential DHT stage, with 75 ohm resistors from each leg (+ and -) to earth before the grid stoppers. That's astounding, and not too expensive at all.For the DHT think 12A, 01A, 26 globe, any of those. One of the least microphonic is probably the VT-67. You need a very dead chassis that doesn't ring at all, but if you use DHTs in differential rather than SE you should be fine. Transformer out gives you the choice of SE or balanced to the amp. If you have a push pull amp it's useful to change it to balanced operation and use a shunt attenuator which you can easily build yourself with really good quality resistors. Kevin Carter can help you out with all this on the K&K forum - he has a 26 line stage and information on all the right Lundahls, plus of course the RAKK DAC which is extremely detailed and clean sounding - a real joy.
Follow Ups:
Any suggestions for what output transformer to use for differential 26 stage?Sounds very tempting
I'm using a LL1660 which seems to work fine. LL1689 works fine as a stepdown. That's my experience.
Andy, because Rune asked about Aikido, I would add that I have built an Aikido stage using DHTs without resorting to DC on the filaments. AC hum on the filaments of the DHTs cancels in the Aikido circuit. In my implementation the microphonics were under control too, which I believe to be another cancellation effect.AC on filaments should be easier to implement than the various DC ideas. However, there is easier and there is easy -- the rub is that you need separate AC filament sources for each DHT in the circuit (I have used separate hum pots with each, too). My experiments have been with 26s, for which four 1.5v sources are needed. I have self wound a little toroid to see if there is a cheap and relatively easy way to get four separate 1.5v power sources. Not yet tested it in circuit.
Rune asked about the sound. Like any proud papa, I like my invention, and have listened to it daily as a headphone amp for five months. But, I cannot offer a comparison of anything else to it, other than my earlier attempts. It is better than any my earlier implementations of headphone amps.
Thanks alot Andy for sharing!I have no electronic background besides I've built some Audio Note kits from good build descriptions,so I can use a solder iron and do some simple measuring:-) ,but besides that most of the technical aspects are totally new ground for me.But I've found out that there is only one way to really learn and thats by practice,and I also learn a lot from friendly guys like you that are willing to share! :-)
There is one thing if you don't mind...I've seen alot of talk about current sources and I have done some quick search and reading,but I didn't understand it.Can you in an easy way explain what it is? Or some good links for me to read about it? This might be a stupid observation/statement but I thought that all power supplies deliver current......?
A standard power supply is voltage regulated; that is, a sample of the output voltage is placed into an error amplifier, and compared to a fixed reference. The error amp determines if the voltage needs to increase or decrease, and will control the regulating device (tube, xistor) accordingly.A current source is essentially the same, except that a sample of the output current is placed into the error amp (this is typically done by dropping the current across a stable resistor to obtain a proportional voltage).
End result is that with a varying load, a voltage source will keep the voltage across the load constant (and current will vary), and a current source will keep the current through the load constant (and voltage will vary).
There are various theories about which is preferred for filaments, but essentially the power supply topology determines if the filament is receiving constant voltage or current.
Does this mean that one have to choose voltage or current regulated,or is it possible to feed the circuit with both ?
So this power supply showing for the Karna ; http://www.nutshellhifi.com/Karna.gif ,is voltage regulated only?
You can only have one or the other. You could have a current LIMITED voltage regulator, but it's still voltage regulated. If your current output (which will vary in a VR supply) goes too high past a design spec, the output will drop, effectively limiting the current output.The Karna design uses shunt voltage regulation for the B+ driver stages, but does not regulate the filaments. It basically is a voltage supply, with regulation determined by the 120VAC of your house power. If your house power drops to 115V, all filament supplies will drop by the same percentage. So while it IS a voltage supply, it is not specifically regulated.
Technically, it has pretty good load regulation, but no line regulation. That is, as the load varies, the voltage will be fairly constant. As the line (house power) varies, the voltage will change.
FWIW, I am currently building a Raven preamp, and am debating between an Amity or Karna design. I can say this much, as a licensed PE with good knowledge in both power and electronic theory, the Karna is a VERY daunting design. It may look easy and simple on paper, but the actual build and attention to subtle grounding is critical. If you are having some of the questions you are posing, I would kindly and respectfully steer you away from building it. At a minimum, try a "tiny" Raven design first, to see how successful you are.
I will consider the build of Amity/Karna,but as you said it also look managable from looking at schematics.I just figured since it has been around for a while I could get help if I ran into problems,but I will now consider not building it.
Maybe a Raven,yes.
> Does this mean that one have to choose voltage or current regulatedThe vast majority of regulated filament supplies are voltage (not current) regulated. One good reason for this is that the filament supply usually feeds a number of tubes in parallel. With a current source, removing a tube (or having a filament open unexpectedly) will cause the supply to increase the voltage to all the other tubes. So, safe operation requires either a crowbar across the output of the current regulator or a separate current regulator for each tube. On the other hand, the downside to voltage regulators (and simple filament transformers) is the inrush current when the tube is cold. This can be minimized if steps are taken during the design of the regulator to provide a slower start.
One solution to supplying two tubes - e.g. a diff pair - with one current source is to wire them in series. So if you remove one tube you break the circuit.
> ...wire them in series.Yes, that's a good option, although it does make it more difficult to know which tube of the pair has failed. I have to say, my preference in all this is a voltage regulator (voltage source) with a current limiter, not merely a simple current regulator. I believe tubes are spec'd and designed for a specific filament voltage, and that the filament current listed for most is only an approximation provided for use in determining supply requirements. Regulating the tube to that current is unlikely to be what the manufacturer intended.
my preference in all this is a voltage regulator (voltage source) with a current limiter, not merely a simple current regulator. I believe tubes are spec'd and designed for a specific filament voltage, and that the filament current listed for most is only an approximation provided for use in determining supply requirements. Regulating the tube to that current is unlikely to be what the manufacturer intended.I used to think that, I even wrote an article in AudioEXpress on the topic.
After adding series chokes to my battery supplied filaments I'm thinking the current reg has benefits above and beyond lighting the things correctly. It looks as though extra impedance in the supply line is beneficial - whether it blocks hash getting in or out I can't say, but it seems to do something.
BTW you can get around the issue with current v voltage by making the current reg adjustable and setting it so that the voltage drop across the filament is correct.
I agree with those who say that the high impedance keeps the audio signal out of the power supply. The last thing i want is a good portion of my audio signal finding its way to ground though the filament supply.How did you size your chokes for your battery supply? I pretty much size it so the Z at 20hz matches the filament resistance and see what that looks like.
For the 71As I used a pair of existing chokes which were about 1H and had about the right DCR to drop nominal 6.3V to 5V. I added a common mode toroidal which tops up the DCR.For the 211s I bought some big transformers off the shelf, unstacked it, unwound the primary and added a new winding. The winding roughly matches the DCR of the existing secondary to give me almost 0.8 ohms as required to drop nominal 12.6 to 10V with the 3.25A draw of the 211 filament. I will add an air gap with paper between the Es and Is and put it on the DCR bridge to see what I get and use the same trick with the common mode toroidals as above.
I have only just sorted the recharging for the 211 batteries so they haven't been installed yet. I'm using laptop power supplies (15V) and a silicon bridge rectifier to achieve 13.8V, the laptop supplies are current limited at 6A so this should work a treat for the 65 Ah batteries.
I'm curious, being as you're using DC on the filaments, why aren't you using SS devices for isolation? This seems like a very inconvenient application for chokes given the frequencies and currents involved. I should probably add that I'm on the outside of the current rage for SS CCSs, but this particular use would seem much less likely to color the sound. The impedance ratio of desired to undesired signal paths would be extremely high with a SS CCS.
The original circuits used fairly high spec SS devices (LT108X series regs). With the 71As I moved from there to series / shunt regulation using discrete devices, from there to SLA batteries and from there to SLA batteries plus chokes. I felt that each step was a worthwhile improvement. With the 211s I'm in the process of jumping straight from first to last.I tried SS current reg but didn't like it, YMMV etc etc.
I tried SS current reg but didn't like itYou probably have some theory about the reason why. No?
I never tried batteries, probably should give them a go.
This post is made possible by the generous support of people like you and our sponsors: