Progress is slow and sporadic right now. Sherri's mom has cancer and was hospitalized this weekend (1200 miles from here). We spent the weekend preparing for her to go stay with her for a while (again). The company that I work for seems to excel only at changing their mind and laying off engineers. This makes for some long and frustrating work days. I have 4 TubelabSE's and a box full of mosfets occupying my entire work space right now, so the P-P stuff is on the shelf right now. I hope to get back to it in the next two weeks or so. I can't even guess when it will be ready.
You are right, one does not require the other, but both are requirements for this design, as is worldwide parts availability. I chose the 6AQ5 / 6005 / EL90 for the output tube because it fulfills these requirements. This amp will be built by people of all skill levels, so simple and easy are also needed, as well as a very high probability of success on first power up (no adjustments).
That might be a little over the top, but I have considered a LTP with 6AU6's or other similar pentodes.
I understand the advantages of a P-P class A amp. The only P-P amp that I currently own runs 300B's in class A P-P. It ROCKS! There is nothing to keep the end user from setting the amp up for class A operation. In my usual style I will probably test every possible combination (triode, UL, pentode, Class A, Class AB1) and publish it all on the web site.
What if any reason that IGBTs have not been considered for
grid superdrives? They certainly seem good for the voltage.
I don't think fully saturated switching speed is an issue for
non-saturating audio circuits, and Thyristor lockup effect is
said to be a solved problem of the past...
How low must miller effect be to satisfy your requirement?
grid superdrives? They certainly seem good for the voltage.
I don't think fully saturated switching speed is an issue for
non-saturating audio circuits, and Thyristor lockup effect is
said to be a solved problem of the past...
How low must miller effect be to satisfy your requirement?
At the time I was designing the PowerDrive circuit (about 5 years ago), the only IGBT's that I could get were large capacitive monsters made for motor control circuits. So, no I have not tried them.
Mr. Miller was not the reason for PowerDrive. The main reason for PowerDrive is to provide a low impedance (less than 100 ohms) to drive the grid, and to be able to supply any grid current that the output tube may draw whether intentionally or due to overdrive.
In either case, I don't plan on using PowerDrive in the SimpleP-P. It takes away from the "simple" part.
Re: Miller Time
Miller, I thought similar or same as reverse transfer?
If not, then what is the actual significant difference?
------------------------------------------------------------------
Google yerself up an "stgp10nc60hd" IGBT
vs your current favorite 2SK3563 MOSFET
600V C-E
500V D-S
+-20 Gate-Emitter
+-30 Gate-Source
365pF Gate
550pF Gate
8.3pF Reverse Transfer
7pF Reverse Transfer
65W
35W
1.9 Volts VCE on
1.35 Ohms RDS on
Hard to say if there is a clear winner for your purpose here today.
I am sure there are better IGBTs, my survey wasn't very thorough.
At the time of my posting, Mouser had 1,000 in stock.
$1.54 by the onezies... TO-220
Miller, I thought similar or same as reverse transfer?
If not, then what is the actual significant difference?
------------------------------------------------------------------
Google yerself up an "stgp10nc60hd" IGBT
vs your current favorite 2SK3563 MOSFET
600V C-E
500V D-S
+-20 Gate-Emitter
+-30 Gate-Source
365pF Gate
550pF Gate
8.3pF Reverse Transfer
7pF Reverse Transfer
65W
35W
1.9 Volts VCE on
1.35 Ohms RDS on
Hard to say if there is a clear winner for your purpose here today.
I am sure there are better IGBTs, my survey wasn't very thorough.
At the time of my posting, Mouser had 1,000 in stock.
$1.54 by the onezies... TO-220
7199
A bit OT, but can someone explain why the 7199 got such a bad name that all and sundry had to design a better driver for the Dyna ST-70?
From the http://www.triodeel.com/7199.htm above it seems a high quality and versatile tube. Or have I [as usual] got it completely wrong?
Cheers all....
A bit OT, but can someone explain why the 7199 got such a bad name that all and sundry had to design a better driver for the Dyna ST-70?
From the http://www.triodeel.com/7199.htm above it seems a high quality and versatile tube. Or have I [as usual] got it completely wrong?
Cheers all....
Hi George,
I have used a variation of this circuit before, with a CF instead of the concertina.
By exchanging the two 100k resistors with a 200k potentiometer you can vary the gain from purely pentode- with the wiper at the 1uF capacitor, to pseudotriode-mode with the wiper at the triodes cathode. To get the best result I used a revlog.
I have used a variation of this circuit before, with a CF instead of the concertina.
By exchanging the two 100k resistors with a 200k potentiometer you can vary the gain from purely pentode- with the wiper at the 1uF capacitor, to pseudotriode-mode with the wiper at the triodes cathode. To get the best result I used a revlog.
Hi again,
To get equal gain from the concertina you have to add a resistor of the same value as the feedback resistors/potentimeter. Add it after the capacitor on the triodes anode and to ground. When I simmed with a EF86 and a 12AX7 I also had to parallell it with a small cap of 10p.
To get equal gain from the concertina you have to add a resistor of the same value as the feedback resistors/potentimeter. Add it after the capacitor on the triodes anode and to ground. When I simmed with a EF86 and a 12AX7 I also had to parallell it with a small cap of 10p.
This is just my opinion, but there are several factors for the 7199 being designed out of the ST70.
First, the 7199 was getting scarce and expensive starting in the 80's. A NOS 7199 today costs more than I paid for the amp used in 1973. LOL
Second, there is a certain amount of fashion in circuit design. Neither the pentode or the cathodyne phase splitter were /are fashionable.
Third, the phase splitter was marginal when driving an EL34 in triode. (Works well in the original UL configuration)
Forth, while the power supply was the major limitation of the design, the front end was the easiest thing to modify while keeping the same chassis.
Doug
"ohh ohh ohh mr Kotta! irgs4b60kd1pbf"
Seeing as Fairchild came up with the FQP1N50, I looked to see what they had for IGBTs and found this:
SGR2N60UFD 600V, 25W, 98 pF Cinput, 18 pF Coutput, 4 pF C reverse (and a quite flat curve), $0.659/100 D-pak (Mouser)
Don
Seeing as Fairchild came up with the FQP1N50, I looked to see what they had for IGBTs and found this:
SGR2N60UFD 600V, 25W, 98 pF Cinput, 18 pF Coutput, 4 pF C reverse (and a quite flat curve), $0.659/100 D-pak (Mouser)
Don
Best spec so far IGBT or MOSFET, near as I can tell.
Of course you have selected a package that has to be
soldered to a heatsink.... Most likely the board itself...
Is there a through hole version?
Of course you have selected a package that has to be
soldered to a heatsink.... Most likely the board itself...
Is there a through hole version?
7199 fate clarified.....
Thanks for that, Doug, all is clear now--much appreciated.
Ken.
DougL said:
Thanks for that, Doug, all is clear now--much appreciated.
Ken.
"Of course you have selected a package that has to be
soldered to a heatsink.... Most likely the board itself...
Is there a through hole version?"
Not that I see listed at Mouser, unfortunately. I could settle for an I-pak case (coll./drain lead), but nothing there either. TO-220 seems to be a fading package these days. D-Pak ssssucks, I agree. (no coll. lead)
I don't see how they expect us to dissipate 25 watts thru the PC board, but how else to get a collector connection? Guess one would have to use a copper foil heatsink to solder it to and also to serve for the collector connection. Wonder if anyone sells the little copper thingies used for makiing TO-220 xsistors, just solder one on. (I can just see it now, 10 years from now, people will be selling NOS TO-220s, or raiding junk yards to pull the copper tabs off old ones. )
Don
soldered to a heatsink.... Most likely the board itself...
Is there a through hole version?"
Not that I see listed at Mouser, unfortunately. I could settle for an I-pak case (coll./drain lead), but nothing there either. TO-220 seems to be a fading package these days. D-Pak ssssucks, I agree. (no coll. lead)
I don't see how they expect us to dissipate 25 watts thru the PC board, but how else to get a collector connection? Guess one would have to use a copper foil heatsink to solder it to and also to serve for the collector connection. Wonder if anyone sells the little copper thingies used for makiing TO-220 xsistors, just solder one on. (I can just see it now, 10 years from now, people will be selling NOS TO-220s, or raiding junk yards to pull the copper tabs off old ones. )
Don
All the Class-D amps I work with - heatsink though the board.
40Watts per channel, 80% efficient. Whats that, like 16Watts?
And we are not talking particularly large boards...
Any re-soldering that sort of heatsink PCB is a serious pain...
You need a heat gun and an iron (or two) and four hands...
And a whole lot of flux...
Big center pad hidden up under the chip is the real challenge.
Any part where you can at least get an iron on the main pad
should be floatable by the average man with some genuine
leaded Kester... We don't get to use.
40Watts per channel, 80% efficient. Whats that, like 16Watts?
And we are not talking particularly large boards...
Any re-soldering that sort of heatsink PCB is a serious pain...
You need a heat gun and an iron (or two) and four hands...
And a whole lot of flux...
Big center pad hidden up under the chip is the real challenge.
Any part where you can at least get an iron on the main pad
should be floatable by the average man with some genuine
leaded Kester... We don't get to use.
The D-pak case does have a small metal tab that sticks out the top that you can get an iron on. But the short leads and lack of a collector lead are close to show stoppers.
Only semi-practical setup I can think of would be to solder a 1 inch square copper sheet to the collector tab and then use an adhesive backed heatsink, made for uP chips, to push on to that. Lots of those kind of heatsinks available at the surplus outlets. Or solder on a 1 inch square copper sheet with a mounting hole punched in it, like TO-220.
But the short leads, I only see messy solutions there.... If it only had the full three leads, one could probably find a push on Molex pre-wired fan connector.
Only semi-practical setup I can think of would be to solder a 1 inch square copper sheet to the collector tab and then use an adhesive backed heatsink, made for uP chips, to push on to that. Lots of those kind of heatsinks available at the surplus outlets. Or solder on a 1 inch square copper sheet with a mounting hole punched in it, like TO-220.
But the short leads, I only see messy solutions there.... If it only had the full three leads, one could probably find a push on Molex pre-wired fan connector.
I'm sorry, I thought you were referring to the Miller capacitance of the output tube.
In the traditional sense Miller capacitance is the grid (gate) to plate (drain) capacitance multiplied by the device gain in the chosen circuit. I don't think that the "reverse transfer capacitance" spec considers the gain. It doesn't matter since the plate (drain) is at AC ground in a follower application. The grid (gate) to cathode (source) capacitance is bootstrapped since the cathode follows the grid, and is nearly at the same AC potential. It would appear that the driver tube sees the "reverse transfer capacitance" and stray capacitance across its output. Obviously the choice of driver tube, and its operating conditions determine how much capacitance is allowable. A 12AX7 is not going to handle as much capacitance as a 45.
It has been pointed out to me by people much smarter than myself that all semiconductors act as voltage variable capacitors. The "reverse transfer capacitance" varies with applied voltage. It is believed that this variable capacitance can cause "phase intermodulation distortion". Of course PIM is very hard to measure, and it is also one of the reasons for the belief that "all sand in the signal path is evil".
This effect is lessened as the voltage is raised, so for small amps like the TubelabSE the solution is to run the fet at full B+ voltage. It is also best to choose a fet with a flat capacitance VS voltage curve.
OK, I'll get some the next time I order some parts. At the very least I will have a new kind of part to blow up! SMD is out of the question for devices that may wind up on one of my commercial PC boards. Many of my users have never assembled a PC board before, and can't deal with SMD parts. I have no problems with anything bigger than 0204 parts. My eyes are to old and my hands are too shakey for 0204. We even use 0102 at work, and now they have invented 00501's.
It does seem that the TO220 is a dying breed. Some manufacturers have already killed them off.
I work on RF power amps for two - way radios. We must pass a 7 minute key down at 100% duty cycle test. Any more than 5 watts through the board (on top of solder filled vias to a heat sink) can cause the device to reflow. I have also managed to unsolder the PA device on some cell phones by running them in test mode at 100% duty cycle.
80% must be worst case pessimistic spec then... Never even got
one too warm to touch, not unless something else was seriously
wrong.
But you can't flow these parts out of place with a pair of 40
watt irons unless you got a heatgun or two helping... Ground
plane on these boards is pretty substantial.
I havn't done this sort of work for more than 15 years, I don't
know what is normal anymore? This isn't soldering the way I
learned it from Heathkit.
I let the girls with the tiny hands do most of the microscope
work if these is no time pressure. I don't get involved lessn I
got a problem I need to keep my train of thought focused on
one specific board.
My SMT methods are completely feral at best. Blob it on and
clean it up with wick... I don't pretend to have any hand-eye
coordination.
-----------------------------------------------------------------------------
As for the Fairchild IGBT, jut need a through hole big enough
for 1/2 inch copper pipe... Problem solved!
one too warm to touch, not unless something else was seriously
wrong.
But you can't flow these parts out of place with a pair of 40
watt irons unless you got a heatgun or two helping... Ground
plane on these boards is pretty substantial.
I havn't done this sort of work for more than 15 years, I don't
know what is normal anymore? This isn't soldering the way I
learned it from Heathkit.
I let the girls with the tiny hands do most of the microscope
work if these is no time pressure. I don't get involved lessn I
got a problem I need to keep my train of thought focused on
one specific board.
My SMT methods are completely feral at best. Blob it on and
clean it up with wick... I don't pretend to have any hand-eye
coordination.
-----------------------------------------------------------------------------
As for the Fairchild IGBT, jut need a through hole big enough
for 1/2 inch copper pipe... Problem solved!
"just need a through hole big enough
for 1/2 inch copper pipe... Problem solved!"
We just need a brass or copper hex spacer similar to this one:
http://www.goldmine-elec-products.com/prodinfo.asp?number=G15351
Solder the IGBT to one of the flats. Put an aluminum heatsink on the top with a bolt thru it and thru the spacer into the PC board for mounting and for the drain connection. The IGBT pins go into the PC board for soldering.
Best assembly method would be to solder the IGBT pins into the PC board, then temporarily mount the copper spacer with a bolt, and solder the two together with a flat side against the IGBT. Then loosen the bolt and add the aluminum heatsink on top. Re-tighten the bolt.
Here is one place that makes custom spacers:
http://www.accuratescrew.com/
I've requested a quote for 100 copper pcs. I will report back when I know how much. Most likely other places have something suitable off the shelf, likely brass ones.
Don
for 1/2 inch copper pipe... Problem solved!"
We just need a brass or copper hex spacer similar to this one:
http://www.goldmine-elec-products.com/prodinfo.asp?number=G15351
Solder the IGBT to one of the flats. Put an aluminum heatsink on the top with a bolt thru it and thru the spacer into the PC board for mounting and for the drain connection. The IGBT pins go into the PC board for soldering.
Best assembly method would be to solder the IGBT pins into the PC board, then temporarily mount the copper spacer with a bolt, and solder the two together with a flat side against the IGBT. Then loosen the bolt and add the aluminum heatsink on top. Re-tighten the bolt.
Here is one place that makes custom spacers:
http://www.accuratescrew.com/
I've requested a quote for 100 copper pcs. I will report back when I know how much. Most likely other places have something suitable off the shelf, likely brass ones.
Don
Considering the amount of time and heat needed, I would
heat the hex with a heatgun or propane torch first, melt on
some solder, then apply the IGBT. Only after that cools would
I bolt that to the PCB to solder the last two leads.
But the entire excercise is rediculous unless no suitable TO220
part exists. I just don't see that as an immediate problem for
the circuit at hand.
heat the hex with a heatgun or propane torch first, melt on
some solder, then apply the IGBT. Only after that cools would
I bolt that to the PCB to solder the last two leads.
But the entire excercise is rediculous unless no suitable TO220
part exists. I just don't see that as an immediate problem for
the circuit at hand.
- Status
- Not open for further replies.