I strongly suspect that the 68ohm resistor in the NewClassD soft starts is sized for European 230V systems. Most references I can find recommend more like 20-30ohm for 120V systems, so I've ordered some 22ohm, 15W resistors to swap in, and expect that will alleviate the inrush issue on the 833. I'll report back once I get them.
Yup, Lars from NewClassD just confirmed that the 68ohms is optimized for 230V. He also offered to reprogram the controller chip if need be (I'll try the resistor first). Very good service from them!
Status Update:
The startup saga continues...I received the 22R, 14W resistors and installed them in the soft start board for the 833, but it made no difference. The 10A fuse still blows unless the 833 bias voltage is increased to drop the current to ~50-80mA on startup; after startup it can be increased to 150mA or more with no issues. I think I just need a slightly bigger fuse, so I'll try a 12A. I'm using Radio Shack slo-blo fuses, which honestly don't look any different from their fast-blo versions, with a single thin element...maybe I'll look for some more robust slo-blos, with the coiled elements.
Today I did an extended listening session after my wife went out for dinner with the girls; just me sitting about 4 feet from the speaker running through some of my favorite CDs. Three hours plus, and I have to say, this amp sounds EXCELLENT. Very, very detailed, clean and clear, with very strong, yet well-defined bass, like a GIANT 300B or similar DHT amp with a turbo! The power seems limitless; I had it cranked up LOUD and it did not seem to bother it a bit. I did a quick frequency response measurement with my laptop and pink noise input and it was pretty much a flat line from 20Hz to 20kHz. The laptop was giving me a lot of background noise (-30dB level) for some reason, even with the amp disconnected, but once the signal was applied the output was very flat over the audio band.
The 833 current was set to 140mA but there were a few times when it drifted down to 115mA or so for a few minutes, then went back up to 140mA...peculiar. I was worried that my bias voltages were drifting, but when I measured them they were still rock solid within +/-.3V or so, despite the drifting 833 current. Then I measured my line voltage and found it drifted by 2-3V or so over the course of an hour - that translates to 40-60V B+, which together with the slight bias voltage fluctuations would account for the drift, according to LTSpice modeling. I may have to live with that since the 833B+ is unregulated yet the bias supplies are regulated - anyone know of a good 2300V regulator???!!! I may just need to regulate the wall voltage a bit to stabilize it, with a conditioner of some sort.
Still hunting down the source of the little overshoot and ringing on the front of the 2kHz square wave. So far I know it's not present up to and including the grid of the 833 - the squares look perfect up to there.
.
The startup saga continues...I received the 22R, 14W resistors and installed them in the soft start board for the 833, but it made no difference. The 10A fuse still blows unless the 833 bias voltage is increased to drop the current to ~50-80mA on startup; after startup it can be increased to 150mA or more with no issues. I think I just need a slightly bigger fuse, so I'll try a 12A. I'm using Radio Shack slo-blo fuses, which honestly don't look any different from their fast-blo versions, with a single thin element...maybe I'll look for some more robust slo-blos, with the coiled elements.
Today I did an extended listening session after my wife went out for dinner with the girls; just me sitting about 4 feet from the speaker running through some of my favorite CDs. Three hours plus, and I have to say, this amp sounds EXCELLENT. Very, very detailed, clean and clear, with very strong, yet well-defined bass, like a GIANT 300B or similar DHT amp with a turbo! The power seems limitless; I had it cranked up LOUD and it did not seem to bother it a bit. I did a quick frequency response measurement with my laptop and pink noise input and it was pretty much a flat line from 20Hz to 20kHz. The laptop was giving me a lot of background noise (-30dB level) for some reason, even with the amp disconnected, but once the signal was applied the output was very flat over the audio band.
The 833 current was set to 140mA but there were a few times when it drifted down to 115mA or so for a few minutes, then went back up to 140mA...peculiar. I was worried that my bias voltages were drifting, but when I measured them they were still rock solid within +/-.3V or so, despite the drifting 833 current. Then I measured my line voltage and found it drifted by 2-3V or so over the course of an hour - that translates to 40-60V B+, which together with the slight bias voltage fluctuations would account for the drift, according to LTSpice modeling. I may have to live with that since the 833B+ is unregulated yet the bias supplies are regulated - anyone know of a good 2300V regulator???!!! I may just need to regulate the wall voltage a bit to stabilize it, with a conditioner of some sort.
Still hunting down the source of the little overshoot and ringing on the front of the 2kHz square wave. So far I know it's not present up to and including the grid of the 833 - the squares look perfect up to there.
.
Last edited:
Oh, I forgot to mention temperatures. Over the course of 3 hours of use, the temp of the chassis rose to about 59C at its hottest points (where the filament and shunt regulator heatsinks attach to the top plate and side, respectively). That's probably OK, but I'm thinking of adding an auxiliary heat sink to the top plate behind the 833 tubes just to be safe; I have a pair of nice black anodized ones I can use for that. So, the temp seems to be well handled by the existing scheme.
By the way, the Landfall Systems aluminum chassis is proving to be very sturdy, even with the massive weight of this beast. I can flip it over and back up without any detectable flexing at all; it feels like a solid block. Highly recommended!
By the way, the Landfall Systems aluminum chassis is proving to be very sturdy, even with the massive weight of this beast. I can flip it over and back up without any detectable flexing at all; it feels like a solid block. Highly recommended!
One last thing...I thought that perhaps the stray capacitance of the GDT across the OPT primary might be causing the overshoot seen, so I took it out, but that did not resolve the issue, so it goes back in tomorrow for OPT protection.
More experimenting tomorrow. As I said above, the amp sounds great as is, but I'd like to know where that overshoot/ringing is coming from anyway, even if I don't bother to fix it.
More experimenting tomorrow. As I said above, the amp sounds great as is, but I'd like to know where that overshoot/ringing is coming from anyway, even if I don't bother to fix it.
Hi Magz,
I am trying it in my bad english...
Congratulations!!
The Amp looks very nice!
It is very difficult to find some issues without a complete shematic....
I was reading back the comlete thread...but for the B+ for the 833 I didn't found all...espacally how many sive capacity You use for it. As I said, an empty capacitor is like a short circuit.
May be, You can give it more time for charging, before You Switch on the main Supply directly. But You can also try to use some resistors (about 50ohms/20W) between the main transformer and the first capacitor. (In SE-circuits the last capacitor of the sive makes the impedance of the complete HV-supply.) That's why You can try 100ohms also (but no much more!)...
My opinion:
It is very simple...
You have spend a lot of Money for the OPT...
May be, this are the best OPT's, You can get for money on this planet...
BUT, this are transformers only:
There are some physical laws, you can't remove by money...
- we have a Rdc of the primary of the OPT (between 200 and 400 ohms)
- we have a Rdc of the secondary of the OPT (0.5ohms)
- we have a stray capacitance on both (some pF)
- we need a lot of inductance on the primary (for bass, LF), but this makes a lot of stray capacity...
- we have 12000ohms primary impedance...difficult (or impossible) to wind this witout stray inductance and capacitance...
If You feed the amp with square waves, You are generate a lot of THD (higher order! At the input!)...The result is overshoot and ringing...
But...in normal audio, are squares not included...
Thats why, relax...as You wrote:
Best Regards, Matthias
I am trying it in my bad english...
Congratulations!!
The Amp looks very nice!
It is very difficult to find some issues without a complete shematic....
I was reading back the comlete thread...but for the B+ for the 833 I didn't found all...espacally how many sive capacity You use for it. As I said, an empty capacitor is like a short circuit.
May be, You can give it more time for charging, before You Switch on the main Supply directly. But You can also try to use some resistors (about 50ohms/20W) between the main transformer and the first capacitor. (In SE-circuits the last capacitor of the sive makes the impedance of the complete HV-supply.) That's why You can try 100ohms also (but no much more!)...
Still hunting down the source of the little overshoot and ringing on the front of the 2kHz square wave. So far I know it's not present up to and including the grid of the 833 - the squares look perfect up to there.
My opinion:
It is very simple...
You have spend a lot of Money for the OPT...
May be, this are the best OPT's, You can get for money on this planet...
BUT, this are transformers only:
There are some physical laws, you can't remove by money...
- we have a Rdc of the primary of the OPT (between 200 and 400 ohms)
- we have a Rdc of the secondary of the OPT (0.5ohms)
- we have a stray capacitance on both (some pF)
- we need a lot of inductance on the primary (for bass, LF), but this makes a lot of stray capacity...
- we have 12000ohms primary impedance...difficult (or impossible) to wind this witout stray inductance and capacitance...
If You feed the amp with square waves, You are generate a lot of THD (higher order! At the input!)...The result is overshoot and ringing...
But...in normal audio, are squares not included...
Thats why, relax...as You wrote:
...As I said above, the amp sounds great ...
Best Regards, Matthias
Hi Magz,
I am trying it in my bad english...
Congratulations!!
The Amp looks very nice!
It is very difficult to find some issues without a complete shematic....
I was reading back the comlete thread...but for the B+ for the 833 I didn't found all...espacally how many sive capacity You use for it. As I said, an empty capacitor is like a short circuit.
May be, You can give it more time for charging, before You Switch on the main Supply directly. But You can also try to use some resistors (about 50ohms/20W) between the main transformer and the first capacitor. (In SE-circuits the last capacitor of the sive makes the impedance of the complete HV-supply.) That's why You can try 100ohms also (but no much more!)...
My opinion:
It is very simple...
You have spend a lot of Money for the OPT...
May be, this are the best OPT's, You can get for money on this planet...
BUT, this are transformers only:
There are some physical laws, you can't remove by money...
- we have a Rdc of the primary of the OPT (between 200 and 400 ohms)
- we have a Rdc of the secondary of the OPT (0.5ohms)
- we have a stray capacitance on both (some pF)
- we need a lot of inductance on the primary (for bass, LF), but this makes a lot of stray capacity...
- we have 12000ohms primary impedance...difficult (or impossible) to wind this witout stray inductance and capacitance...
If You feed the amp with square waves, You are generate a lot of THD (higher order! At the input!)...The result is overshoot and ringing...
But...in normal audio, are squares not included...
Thats why, relax...as You wrote:
Best Regards, Matthias
Thanks for the reply, Matthias. The PSUDs for the power supplies are on pages 5-6 of this thread. All choke input.
I'm not worried about the overshoot, just want to know for sure where it is coming from. It may not be from the OPT.
This amp has an interesting design in that the cathode bias voltage is provided not by a resistor, but by the shunt regulator pass MOSFET sinking the 833 current; all the 833 anode current plus grid current goes through that MOSFET on the way to ground, and the regulator circuit keeps the cathode bias voltage steady at approximately 247VDC - in effect the regulator acts like a bypassed cathode resistor, without the big bypass cap. I'm thinking that perhaps the response of the shunt regulator circuit in the cathode of the 833 is responsible for the small amount of overshoot and ringing on square waves. The regulator is a version of the Salas SSHV regulator, sold by K&K Audio, and modified by me to handle the increased dissipation encountered in this application. Perhaps it needs a little more tweaking, perhaps not...more experiments will tell, I think.
Well, I think we can put the square wave question to bed now. I received an email from Ward at Monolith Magnetics that said:
The square wave is exactly what I calculated. It features sharp edges and
a very well controlled mild overshoot.
So there you have it. All looks as it should. The shunt regulator in the cathode seems to be working just fine! It certainly sounds very, very good!
As soon as I get the noise issue with my soundcard cleared up I'll post a frequency response plot. Today I need to put the GDT back in for OPT protection, try a slightly bigger slo-blo fuse on the mains, and perhaps play around a little with the grounding of the OPT secondary to find the lowest noise connection.
.
The square wave is exactly what I calculated. It features sharp edges and
a very well controlled mild overshoot.
So there you have it. All looks as it should. The shunt regulator in the cathode seems to be working just fine! It certainly sounds very, very good!
As soon as I get the noise issue with my soundcard cleared up I'll post a frequency response plot. Today I need to put the GDT back in for OPT protection, try a slightly bigger slo-blo fuse on the mains, and perhaps play around a little with the grounding of the OPT secondary to find the lowest noise connection.
.
Last edited:
Hi Magz,
I was playing a Little bit with your HV and softstart...
This is 50mA load and 68ohms in softstart
This are 50mA and 22ohms
Now 150mA load on 22Ohms
And I think 10ohm should work with 150mA
Hope, that was helpful...
Best regards, Matthias
I was playing a Little bit with your HV and softstart...
This is 50mA load and 68ohms in softstart
An externally hosted image should be here but it was not working when we last tested it.
This are 50mA and 22ohms
An externally hosted image should be here but it was not working when we last tested it.
Now 150mA load on 22Ohms
An externally hosted image should be here but it was not working when we last tested it.
And I think 10ohm should work with 150mA
An externally hosted image should be here but it was not working when we last tested it.
Hope, that was helpful...
Best regards, Matthias
Thanks Matthias, that was indeed very helpful!
I didn't think to simulate it in Spice, because my Spice skills are rudimentary at best ;-).
I guess I'll need to order a couple more resistors!
Edit: I'll try the 12A slo-blo fuse first, though. If that works, I'll probably stay with that. 12A as shown in your last sim through a 10R resistor is 1440W dissipation - I'm a little worried the 14W or so resistor might explode during the 1 second soft start, and I can't fit anything bigger than that on the board.
I didn't think to simulate it in Spice, because my Spice skills are rudimentary at best ;-).
I guess I'll need to order a couple more resistors!
Edit: I'll try the 12A slo-blo fuse first, though. If that works, I'll probably stay with that. 12A as shown in your last sim through a 10R resistor is 1440W dissipation - I'm a little worried the 14W or so resistor might explode during the 1 second soft start, and I can't fit anything bigger than that on the board.
Last edited:
I guess I'll need to order a couple more resistors!
May be...but You can (should) test with two 22ohms in parallel before...
Magz
Wondering how one of these might work in place of a resistor.
http://www.ametherm.com/datasheetspdf/MS2250004.pdf
Might give a really soft start.
Wondering how one of these might work in place of a resistor.
http://www.ametherm.com/datasheetspdf/MS2250004.pdf
Might give a really soft start.
Hi MelB
It is to simple for HighEnd
this is the better choice for this application:
Ametherm MS32 50006-L ---50 ohm / 6 Amp Inrush Current Limiter Data Sheet
Hard to find parts...thank You!
Wondering how one of these might work in place of a resistor.
http://www.ametherm.com/datasheetspdf/MS2250004.pdf
Might give a really soft start.
It is to simple for HighEnd
this is the better choice for this application:
Ametherm MS32 50006-L ---50 ohm / 6 Amp Inrush Current Limiter Data Sheet
Hard to find parts...thank You!
Magz,
there are 1400W for some milliseconds only...
after a half second it is much lower 500W or so, at one second you have 200W and after two 125W...
You could use this one (mounted on the case and wired):
- Serie RB
The RB250 type is nominal 250W...with an overload of 5times (1250W) for 5 seconds...this schould be very save
Matthias
there are 1400W for some milliseconds only...
after a half second it is much lower 500W or so, at one second you have 200W and after two 125W...
You could use this one (mounted on the case and wired):
- Serie RB
The RB250 type is nominal 250W...with an overload of 5times (1250W) for 5 seconds...this schould be very save
Matthias
I'm a little worried the 14W or so resistor might explode during the 1 second soft start, and I can't fit anything bigger than that on the board.
Then place it off the board... Or as others have suggested, use multiple in parallel. You'll need to look at some resistor data sheets to find some that can handle the peak power. You may find that the Vishay RH-xx series of resistors is the best fit in your application.
Another option to explore would be one of the dedicated inrush limiters. I use the CL-90 in my amps. It's probably way too whimpy for your application, but I'd explore options in that direction. The idea here is to have high resistance on start-up. This resistance gradually decreases as the current limiter heats up. That way you don't get that huge current spike when the resistor is shorted out.
Edit: In other words: I second MelB's suggestion...
~Tom
Magz can you piggyback another 22 ohm on top of the one that is there?
Maybe, but it's really tight space-wise by the leads; the 14W resistors I put in are longer than the 10W that were in there originally, and they just fit between the other components.
It's very tight space-wise in that part of the amp, with two separate soft start boards stacked up, the filament xfrmr, the fan xfrmr and power supply, the 6E5P rectifier board, and all the AC switches, fuses and relays. No way I can fit an off-board component in there either.
A real time delay slo-blo fuse should be able to handle the currents in the 150mA, 22R simulation; I question the "slo-blo-ness" of the Radio Shack fuses I was using, as I said, the are simply thin straight wire, with no blob in the middle even. I need to order some with the supported coiled element...most are specced to handle 4X rated current for 5 seconds.
Does the fuse blow on start-up or does it blow when the soft start relay closes? I would guess the latter. What you're seeing there is basically the transient response of the supply filter. The only way to prevent those current spikes is to release the soft start more gradually. Either by using a soft start component that does this already (see the current limiters above) or by staging the soft start. You could have a series connection of 2-3 resistors and gradually release the soft start. But I'm guessing you don't have room for this...
Another thing: Are you running this amp on its own dedicated lighting circuit? In the US, older houses are typically wired for 15 A per circuit. 20 A in newer houses... You'll probably want to consider running a 12 AWG 20 A circuit for your amps. Just saying...
~Tom
Another thing: Are you running this amp on its own dedicated lighting circuit? In the US, older houses are typically wired for 15 A per circuit. 20 A in newer houses... You'll probably want to consider running a 12 AWG 20 A circuit for your amps. Just saying...
~Tom
Hi Magz,
This is the simulation of my HV supply.
As You can see, it is verry,verry simple and not comareable against Yours...
I like simple Solutions. The inrush current is limited by the two 15ohm resistors and the impedance of the power transformer only.
There is 1kV only for B+ with a load of 200mA...but much more capacitance. So you can't compare...
May be, the 12A fuses do the Job, may be not...You have to test it..
Regards, Matthias
Edit:
@tomchr
Magz wrote:
This is the simulation of my HV supply.
An externally hosted image should be here but it was not working when we last tested it.
As You can see, it is verry,verry simple and not comareable against Yours...
I like simple Solutions. The inrush current is limited by the two 15ohm resistors and the impedance of the power transformer only.
There is 1kV only for B+ with a load of 200mA...but much more capacitance. So you can't compare...
May be, the 12A fuses do the Job, may be not...You have to test it..
Regards, Matthias
Edit:
@tomchr
Magz wrote:
So it can't be the transient Response of the supply filter ...The 10A fuse still blows unless the 833 bias voltage is increased to drop the current to ~50-80mA on startup; after startup it can be increased to 150mA or more with no issues.
Last edited:
- Home
- Amplifiers
- Tubes / Valves
- The Midlife Crisis - My 833C Amp Build