|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
74.220.74.24
Here is a schematic for a 7788 drive 45 SET that I drew up. I've collected all the iron, caps, tubes and two chassis but I have some questions before I warm up the soldering iron.
First, I'm new to the CCS game. I'm planning to get either the C4S boards from Bottlehead or Kevin's cascode MOSFET boards. Leaning towards the latter. Do I need an RC filter section to drop the B+ down before the CCS boards? Or can I adjust both the voltage and the current with them? I'm excited to try out a CCS. Even if it increases the output impedance of the gain stage, I don't think it will be a problem to drive that grid choke.
Second, I've heard that the 7788 is prone to oscillations and want to keep that out of the picture. So I have a 100 ohm plate stopper and a 1000 ohm stoppers on the g1 and g2. Do I need a stopper on the supressor grid? And should g2 and g3 be tied to the anode above or below the anode stopper?
Thanks,
Dan
Follow Ups:
OK, so here is an updated schematic with corrections. Thanks for all the help so far. And I'll gladly take any more comments... or wise cracks too.
Dan
not that it matters that much in this situation due ot the CCS and the large inductance of your grid choke but...The curves for the 7788 you have posted assume a bypassed cathode. Leaving the cathode resistor unbypassed will give you a drastically different set of curves.
I don't have a model for the 7788 but running the curves for a 437A with and without a 91 ohm cathode resistor gave me this. The math says the Rp of the unbypassed situation should be around 6K and the curves match that.
dave
Edits: 09/16/09
Hi Dave,
Thanks for pointing that out. I'm still learning here. I hope you don't mind a question. I understand that without the feedback from a bypass capacitor you should get less gain. However, you made your simulation without a cathode resistor. Does the addition of a bypass capacitor effectively remove the cathode resistor from the circuit in your model?
Dan
Hey,
The feedback comes from the unbypassed cathode resistor and in a perfect world if you "properly" bypass that resistor is is removed from the circuit from the AC perspective. It is only used for DC operation. Of course the world isn't perfect and while measurements will not show it, numerous people have reported vast changes in the sound of between different bypassed cathode resistors.
dave
Hello DanK -The 'preferred' triode connection for pentodes is with G3 connected to the cathode; the way you have it will also work, though.
Here's a couple pictures from Tom Schlagen's presentation at ETF '06 about triode-connected pentodes:
I assumed the typo that Allen pointed out was just that, a hasty drawing error. But it would be one way of listening to the power supply being modulated by music. Could be interesting....
Edits: 09/11/09
Thanks for the note. I'm sure I'll wind up testing the supressor tied to the cathode, but I have seen a lot of people that get better results with it tied to the anode on high Gm tubes. Damir found it worked better with both the D3A and the E180f when coupled to a 300b. Also, Slownlo wound up with it drawn this way in his D3A - 300b schematic. Like Dave says, I'll have to report back the results when I try it out.
Dan
As the curves show, the difference is comparable to two different tubes of the same type; (or a new and a used tube) based on this, one can't really be preferred to the other.
That's why I put preferred in quote marks; the author also notes that both ways are valid.The main difference, (imho) and possibly the reason why much literature says 'G3 to cathode' is the Miller capacitance. In some pentodes wired as triodes this could be substantial. Wiring both grids to the plate will result in more Cg-a than the other way.
As to any other subjective preference, decades ago I tried both on some mic pre's and found the answer is 'it depends', on the circuit, and of course, one's taste.
But you only built this to listen to PS noise anyway, right? ;)
Edits: 09/11/09 09/11/09
yeah, solder and listen needs to be done for this case.
thanks,
dan
man, i've gotta get that schematic corrected soon...
The measured differences here are just that. What we need is 20 people to try it both ways and independently report back their findings. At that point we can call up the science and statistics to try to define the behaviour.
dave
hey-Hey!!!,
The K&K ccs offers a low impedance output terminal, aka mu follower that the others don't. This way the valve actually sees constant current, with load current supplied by the ccs. Very nice...:) No other de-coupling needed btw.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Don't fuss the details, look at his schematic - it WON"T work as drawn.
He's new, so it's obviously just a mistake, but that none of you spotted it leaves me speechless!
Regards, Allen (Vacuum State)
PS: Do you need a hint?
That's just a "typo" kind of mistake.
The OPT will need to go to the plate of the output tube, not the B+.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
nt
..but he asked for feedback, and none of you gave it.
Maybe in his eyes it wasn't a typo - maybe he'd build it as drawn and it wouldn't work.
Regards, Allen (Vacuum State)
!
you're right...
would there have been any damage caused by connecting things as such? Or would it just have sounded like shit?
...and you'd most likely hear NOTHING, unless your speakers are HUGELY sensitive, then you might hear the hum, noise and signal residual on the B+ rail.
Regards, Allen (Vacuum State)
You would hear whatever AC signal was present on the B+ rail. That could be inaudible or ...
the "input" looks like the power source.
Me think he's just using that as a signal source (but then, you knew that already)...
I feel like I'm playing one of those old readers digest visual challenges...
You're dropping 145VDC across the source pin R on the CCS, so you'll be dissing about what?.... 5.1 Watts...?
...to be safe you'll need a 10W R before the CCS to ease some of the dissipation burden off it and drop the VDC to a more tolerable level...
as has been suggested to me before by the the wise mages on this site, a Mills 12W should fit the bill just fine...
slownlo
You'll want a dampening resistor across the grid choke or you'll end up with a resonant peak (perhaps as high as 15 - 20dB depending on the GC) that could cause you all sorts of problems.
What value (L & R) grid choke are you using and are you aiming for fully damped (flat response) or slight bass lift and if so, required frequency and +dB?
Rgds,
Naz
Naz
If it had a plate choke I would agree.
With a CCS plate load, there is not going
to be the peak you are referring to.
I will be a simple 2nd order LC filter.
I would use a .33uFd but a .22uFd is fine.
DanL
Dan,
Dave's right but there is one point that will have a significant effect on lowering the peak and that's the OP impedance of a Pentode which is fairly high. By my calcs it will still exist, possibly in the order of 6db at less than 10Hz depending on how much the choke's inductance reduces at this frequency. Increasing the value of the cap to 0.33uF will lower the resonant frequency and further dampen the circuit by perhaps another 2dB.
Naz
we need to be really careful to keep this thread where the 7788 is triode wired separate from another thread with a similar drawing which used the C3M as a pentode.
If the pentode is the source (as in the other thread), the load on the pentode needs to come from somewhere and whether that load is before or after the cap has a distinct impact on the overall CL resonance behavior.
dave
Sh*t Dave I'm losing it ... gotta slow down. I don't play with Pentodes much but that's no excuse for totally missing the triode connection ... thanks.
But that makes the case for a damping resistor all the more important. With the much lower OP impedance of triode connection we are now back to a major peak, albeit at sub 10Hz, and as you pointed out the CCS does nothing to reduce it.
Naz
the resonance is between the C and the following L and the load on the tube only has an effect of the damping of the resonance.
The parallel combo of the load on the tube and the Rp of the tube determines the damping of resonance. Since the load is usually a magnitude larger than the Rp, the Rp will dictate the behavior of the resonance.
It should be noted that a CCS will effectively give you a source impedance driving the CL filter of the tube. A choke will lower in impedance as frequency goes down (lowering the source impedance) which decreases the damping increasing the magnitude of the peak but by the time you get to that point you are also rolling off any LF info that may excite the resonance.
(hope that made sense)
dave
> > A choke will lower in impedance as frequency goes down (lowering the source impedance) < <
I do get it, chokes are typically rated at a frequency, what is it and by what magnitude would you expect it to change by say 20Hz?
Naz
I was referring to the normal 2piFL (Xl) change in impedance of an ideal inductor. What I think you are referring to is the change in inductance that happens in nonlinear inductors with respect to applied voltage and frequency and honestly that opens a can of worms that would need a new topic to avoid confusing the hell out of everyone.
dave
The BCP-16 Ni grid choke has over 3000H inductnace. I don't remember the DCR exactly, but it's something like 3k ohms. The goal is a flat freq response, but I wouldn't be opposed to a slight bass lift. My altecs could use some help there. I was thinking of testing the circuit with a square wave and looking at the signal on the 45 grid. Then fiddle with resitors across the grid choke to try and damp out any ringing. That said, I've never heard anyone suggest that a resistor was needed across the BCP-16.
Dan
Hi Dan:you wrote;
:::That said, I've never heard anyone suggest that a resistor was needed across the BCP-16.:::
There are two pretty decent LC resonance calculators that you can run your numbers through... voltsec has one and gianluca wrote one as well.
Bottom line... with any rudimentary care your grid choke will work well without a damping resistor in most real world cases\circuits.
Actually... I've played with both Voltsec's and GL's programs at length... it takes a fair amount of "engineering" to get the wild twenty and twenty four db peaks... and it is relatively easy to just change component values which in most cases can move the PROGRAM CALCULATED resonance as much as a decade below the audio band.
Bear in mind that ALL of the programs and spice sims fail to model in or take account of core losses as just one example. None of the programs\sims recognize that the indutive reactance itself at two hertz is going to be very much smaller than say at fifty, sixty, or a hundred hertz where most of the L measurements are traditionally made.
None-the-less, in one such imperfect simulation that I did using voltsec's program I did find quite interesting results.... yep... I could engineer in a huge twenty db something peak at 1.85 hertz... but the peak was extremely narrow.... you basically had to hit just exactly the target value of 1.85 hz... move just a tenth of a db off in either direction and the wild peak was largely gone.
The other thing that would happen if you had some wild twenty some db peaks in your response in the sub-sonic hertz regions... IF you were even able to excite the frequency.... it would quickly (assuming your operating at some X amount of output power to begin with) overload many of the irons in the signal path... (as well as many of your other circuit components) which would then act as a helluva brake on passing the suggested massive peak signals at this extremely low frequency. The "offending inductances" themselves would quickly evaporate as you overloaded the iron components ... and hence deprive the errant resonance of one of it's two necessary components.
But first you have to find a way to excite the 1.85 hertz resonance.... what program material do you have access to? If your going to say that there's power line noise down here... then you probably need to clean up your power supply... assuming your on say fifty or sixty hertz mains.
So basically... your comment that not much trouble has ever, ever been reported in real world applications is just plain utterly true. The horror scenarios seems to be the fodder of the war games players as opposed to many (or any) real world experiences by end users.
And if one did find (in a real world documented example) an untamed resonance impinging on the real world performance of their amp... and a simple change of some of the key component values did not suffice in snuffing out the resonance there are then other alternatives (Voltsec's DrP as just one example) to putting a resistor across the grid choke.
msl
Builder of MagneQuest™ & Peerless™ transformers since 1989
Edits: 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09 09/11/09
I'm not suggesting any real world problems will necessarily exist but it's just plain good design practice to take into account all of the known possibilities and in this instance the addition of a simple resistor seems to me the prudent thing to do.
Also, if your comments were levelled at my post, nowhere did I talk of horror scenarios. In fact I like to use "controlled" peaks to my advantage and have advocated their use for years because they work and to my ears sound better in many circuits / systems for obvious reasons.
You're making assumptions that all people like me who like to design a circuit first in order to predict its POSSIBBLE behaviour never actually build, measure and listen to the thing. IME, in most cases Spice models are pretty accurate compared to the real world, especially in simple examples such as this.
The peaks DO occur and it's plain stupid to completely ignore them. In this case it will be large and unless I'm missing something, in the region of 5 - 7Hz or perhaps a little higher in the real world depending on the choke. Whether it causes a problem is entirely another matter but if it doesn’t add anything why take the risk? On the other hand if it’s understood, intended and controlled in both frequency and amplitude (Q)then I’m saying by all means go for it or at least try it!
Naz
Hi Mike,
Thanks for the confirmation. The OPTs to be used are from your custom build of Robin Hoods that you did last year with bell ends. I can't wait to hear them.
Dan
At this inductance and cap value the resonant frequency will be too low to be useful (sub 10Hz). For a flat response you'll need 100K or so by my reckoning.
It's counterintuitive but it would take a MUCH lower cap value to provide boost at say 30 - 40Hz with your choke and you may not have much power headroom with a 45 anyway. A 300H choke with a 0.1uF cap (even undamped) would help if you want to attempt this.
Naz
Hi Dan,
You adjust the voltage across the CCS using the cathode resistor of the 7788 (set at 125 ohm in your schematic). There are three factors at work here: the current through the triode, which is fixed by the CCS; the cathode bias; and the plate-cathode voltage. By altering the bias, you also alter the plate-cathode voltage.
Regarding stoppers, the way you have connected the g2 stopper to the plate stopper is right, i.e. each stopper should be independent. Do the same thing with the suppressor, maybe using 1.5k as a stopper.
mu on a 7788 is ~50, maybe higher. You have a 125R unbypassed resistor for bias. This adds more than 6K to the output impedance of the stage, which frankly makes using a 7788 a waste of effort and money. Might as well use a 6J5 - cheaper, easier, and will have a similar Zout. Otherwise, replace that resistor with a LED -- some blue ones are in the 3.5V range, use a bypass cap, or find another bias method (hint -- use a CCS for the heater and run the current through the bias resistor).
Second, your numbers don't add up. At 28mA with a 3.5V bias, the 7788 will have ~240V on its plate, not 190V, which means it will be dissipating close to 7W which is too much. I'd tone it down to about 15mA and a -3V bias which will put the plate close to 190V and dissipate less than 3W.
At any rate, 28mA is high here. This tube sounds better with less current, probably because it is more prone to oscillation as the current increases.
100R from G2 to the plate is enough, so long as layout is tight.
A cathode resistor bypass cap in this circuit actually does nothing. Because there is a constant current going into the plate there will be a constant current coming out of the cathode. With a constant current flowing through the resistor there will be a constant voltage across the resistor, hence no AC signal, thus a cap will do nothing.
John S.
Thanks for the comments. I was using the 7788 mainly because I had some nice Amprex versions on hand that I wanted to build with. Good catch on the bias point typo. I was planning to use the recommended points from the B-52 article in Sound Practices 15. Grego thought that this tube sounded best at higher currents and used 190V, -2.5V battery bias (not -3.5), at 28mA. Also Pete Millet showed some nice distortion data operating at 166V and 26mA. I think I'll play around with it to find what sounds best. I was trying to get rid of the bypass cap, but I would probably have enough gain even with it in there. A CCS for the heater sounds interesting.
Dan
You are going to toast out about 4 watts on that CCS, be sure to use an appropriate heatsink. Tie those grids right to the plate, and let the anode stopper be the stopper on all of them.
If you can drop the B+ down to 295-300v, you will lighten the thermal load on your CCS.
Thanks for the response. I was using the 275V, -56V RCA operating points. RCA also has a recommended 250V, -50 point but I never see anyone using that one. Maybe I should give it a try.
Dan
I'm using 45s at that operating point, but into a 3K output. Works fine, but I haven't tried any other operating points yet. I had the trannies before I had the tubes, and 250V,-50V,34mA looked more appropriate for the 3K trannies.
Regards,
John
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: