The correct current rating for a power transformer is a subject of some debate. By the book engineers (*) can quote the maximum current the amp will ever draw and state that the power transformer should be capable of supplying that much with some safety margin. Leo, and many guitar amp makers went to the other extreme........hoe small of a transformer can we get away with and have thins thing outlive the warantee period. Reality should be somewhere in the middle.
The amplifier will spend most of it's time idling. If left alone over a period of hours the transformer and the rest of the amp will reach thermal equilibrium that should be somewhere in the "put your hand on the transformer for several seconds and not use foul language" range. Hammonds run HOT, even without a load.
If the amp were pushed to maximum power and left there it would get hotter. Again the transformer should not be too hot to put your hand on for several seconds.
An audio amp will not usually be run at maximum power constantly until it reaches maximum temperature. Even a heavily thrashed guitar amp spends some time idling and at less than max power in between notes. If the transformer is sized such that it doesn't get too hot under the conditions it will see under the most extreme operating conditions, then it is OK to bend the current ratings on the power transformer on musical peaks.
I have been using Hammond power transformers in guitar amps for nearly 50 years, other than one that got wet, I have never blown one, and I don't just bend the ratings......I break them. They do get hot, and I have one that operates in the swear word range in a class A HiFi amp. One of my workhorse amps pulls over 220 mA continuously from an Allied 6K7VG (made by Hammond) rated for 150 mA. That amp is over 10 years old.
Slap a big fat cap (1000 uF or more) across that one ohm resistor, then set the meter on DC volts and thrash away on the amp to see how high you can get the voltage. This is the average current, and that's what heats up the transformer, not the peaks. Play it the way you intend to use it and you can see what the transformer will need to supply.
I recently build a big HiFi amp and needed to find out how much transformer I will need. The peaks were over 600 mA, but I have decided on 400 mA transformer. The test is here:
YouTube
(*) I am one of those engineers
The amplifier will spend most of it's time idling. If left alone over a period of hours the transformer and the rest of the amp will reach thermal equilibrium that should be somewhere in the "put your hand on the transformer for several seconds and not use foul language" range. Hammonds run HOT, even without a load.
If the amp were pushed to maximum power and left there it would get hotter. Again the transformer should not be too hot to put your hand on for several seconds.
An audio amp will not usually be run at maximum power constantly until it reaches maximum temperature. Even a heavily thrashed guitar amp spends some time idling and at less than max power in between notes. If the transformer is sized such that it doesn't get too hot under the conditions it will see under the most extreme operating conditions, then it is OK to bend the current ratings on the power transformer on musical peaks.
I have been using Hammond power transformers in guitar amps for nearly 50 years, other than one that got wet, I have never blown one, and I don't just bend the ratings......I break them. They do get hot, and I have one that operates in the swear word range in a class A HiFi amp. One of my workhorse amps pulls over 220 mA continuously from an Allied 6K7VG (made by Hammond) rated for 150 mA. That amp is over 10 years old.
Slap a big fat cap (1000 uF or more) across that one ohm resistor, then set the meter on DC volts and thrash away on the amp to see how high you can get the voltage. This is the average current, and that's what heats up the transformer, not the peaks. Play it the way you intend to use it and you can see what the transformer will need to supply.
I recently build a big HiFi amp and needed to find out how much transformer I will need. The peaks were over 600 mA, but I have decided on 400 mA transformer. The test is here:
YouTube
(*) I am one of those engineers
yes and diy'ers are not limited to that...the difference between one traffo and several hundred traffos are several hundreds of dollars...
And the answer to “why is the current draw so much higher than expected?” was screen current. I thought I had measured it at about 8mA but it was off the charts... well, on the charts but damn near as high as the plate current. I ended up putting my scope on the 1 ohm resistor (thanks for the tip, Gnobuddy) and confirming the current that way. After I knew the numbers were solid I rechecked everything.
I liberated a 500v ct 100mA Hammond from a shelved project (now that I know it’s about 1/2 the size it needs to be I don’t feel bad hacking it up). Under full load this one settled to about 315v, 330ish at idle. I moved my screen stoppers up to 6k8 and increase my load to 16k plate to plate. In full overdrive current maxes at 105ma so I’m safe on that front but I’m certainly (still) over dissipating my screens. Had to put it to rest for the day.
Still more to do, thanks for all help so far.
I liberated a 500v ct 100mA Hammond from a shelved project (now that I know it’s about 1/2 the size it needs to be I don’t feel bad hacking it up). Under full load this one settled to about 315v, 330ish at idle. I moved my screen stoppers up to 6k8 and increase my load to 16k plate to plate. In full overdrive current maxes at 105ma so I’m safe on that front but I’m certainly (still) over dissipating my screens. Had to put it to rest for the day.
Still more to do, thanks for all help so far.
the tube dissipation ratings are there as a guide so that your tubes live longer...but tube guitarist do not really care...still there are those who like bigger b+ sags in their tube amps...that is easy, just add resistors in the b+ line, so then at full power you do not exceed plate/screen ratings......
having a largish screen resistor abetts distortions, something you may find to your liking...
having a largish screen resistor abetts distortions, something you may find to your liking...
This morning I made a few more adjustments and started working on my screen limiting.
Most importantly I decided that I'm feeling OK about pushing my new PT a little past it's ratings. Given some of the comments above, and the fact that I am definitely under utilizing the 6.3 an 5v windings, I don't see this as a bad choice. The transformer doesn't get hot. So, I dropped by primary load to 12k8. This raised the peak current a little bit. I'm about 25% over spec at absolute maximum overload conditions and B+ droops to about 315 from 340 at idle. A 10% drop seems fine. This load line does exceed max dissipation a little but I'm more concerned about getting the screen under control.
In summary, I've tried 4 different approaches to the screens:
The large shared resistor was awful with screen sagging down to 80v. A safe option but not an appealing one.
The zener dropper worked well, and kept most of the tone of option one without as much dissipation. I'll need higher a wattage zener or series string as dissipation is over 10W at peak. To the credit of the 1N53xxb series of 5W zeners, mine heated up enough to melt a solder joint but didn't short out! I used a 1k resistor and the voltage shifted from 205 to about 170. Within dissipation limits. If zeners weren't so prone to failure this would be the standout option.
The small(ish) shared resistor method worked better than I expected. As one can surmise it drops twice as much voltage at idle - but not twice as much at full swing - relative to two separate resistors. I started with 5500 ohms and work my way down towards the 3.2W max dissipation. At 3k the resistor dropped 30v off B+ at idle and about 170v at full overdrive (to 140v). That's just shy of 4W of screen dissipation. It sounded natural and I could probably play it a little more dangerously. That Watkins Dominator I referenced above used only a shared 1k5 with voltages similar to mine.
This is definitely one of the fussier areas in amp tweaking due to the interplay between variables and the dynamic affect on tone. All in all, the options that caused less shift in screen voltage from idle to maximum sounded better to me. What I observed today is that screen current isn't really affected much by the series resistance, it stayed between 17 and 25mA per tube across all options. The resistors really only drop the voltage thereby limiting the dissipation. With this in mind it becomes clear why starting with a properly sized screen rail is crucial.
Brian
Most importantly I decided that I'm feeling OK about pushing my new PT a little past it's ratings. Given some of the comments above, and the fact that I am definitely under utilizing the 6.3 an 5v windings, I don't see this as a bad choice. The transformer doesn't get hot. So, I dropped by primary load to 12k8. This raised the peak current a little bit. I'm about 25% over spec at absolute maximum overload conditions and B+ droops to about 315 from 340 at idle. A 10% drop seems fine. This load line does exceed max dissipation a little but I'm more concerned about getting the screen under control.
In summary, I've tried 4 different approaches to the screens:
- Modestly sized resistors in line with each screen (From 1k to 6k8) directly from B+
- One very large resistor (15k) to a filter cap, then directly to the screens (Borrowed this from a Univox schematic)
- One small(ish) resistor shared between each screen
- A 100v series zener followed by a small(ish) shared resistor
The large shared resistor was awful with screen sagging down to 80v. A safe option but not an appealing one.
The zener dropper worked well, and kept most of the tone of option one without as much dissipation. I'll need higher a wattage zener or series string as dissipation is over 10W at peak. To the credit of the 1N53xxb series of 5W zeners, mine heated up enough to melt a solder joint but didn't short out! I used a 1k resistor and the voltage shifted from 205 to about 170. Within dissipation limits. If zeners weren't so prone to failure this would be the standout option.
The small(ish) shared resistor method worked better than I expected. As one can surmise it drops twice as much voltage at idle - but not twice as much at full swing - relative to two separate resistors. I started with 5500 ohms and work my way down towards the 3.2W max dissipation. At 3k the resistor dropped 30v off B+ at idle and about 170v at full overdrive (to 140v). That's just shy of 4W of screen dissipation. It sounded natural and I could probably play it a little more dangerously. That Watkins Dominator I referenced above used only a shared 1k5 with voltages similar to mine.
This is definitely one of the fussier areas in amp tweaking due to the interplay between variables and the dynamic affect on tone. All in all, the options that caused less shift in screen voltage from idle to maximum sounded better to me. What I observed today is that screen current isn't really affected much by the series resistance, it stayed between 17 and 25mA per tube across all options. The resistors really only drop the voltage thereby limiting the dissipation. With this in mind it becomes clear why starting with a properly sized screen rail is crucial.
Brian
Last edited:
Thanks for sharing your results, that might save some of us a lot of tinkering time in the future.
Have you considered using a MOSFET source follower as a crude voltage regulator, in combination with a low-power Zener diode, or even just a voltage divider from B+ to ground? MOSFETs are better than Zeners at dissipating several watts, and if the screen voltage is greater than half B+, there will be less dissipation in the series MOSFET than in a shunt Zener diode.
You can insert a resistor in between B+ and the drain of the MOSFET to limit the maximum screen grid current; until that current limit is reached, the screen grid voltage will stay pretty stable. But when the limit is reached, the MOSFET goes into saturation, and the source voltage falls, meaning the screen grid voltage falls.
Alternately, you could also insert a smaller resistor between source and screen grids. Effectively, you've lowered the screen grid quiescent voltage, as well as reduced the amount of screen voltage "sag" when screen grid current does increase.
I haven't tried this, but it is one of those ideas I have been meaning to try for a while. In my case, the inspiration for the idea was having too much screen grid "sag" in a preamp pentode during overdrive.
-Gnobuddy
P. Millet has many amps done like that, mosfet for screens...i did one too for a 6550 pp amp build....g2 regulated at 350 volts...
I've currently got a little amp with a series Zener diode (36V) as a screen dropper.
Currently class A with very stiff rails, so Vs is pretty rock solid.
It's likely to grow a switch or two to allow a controlled amount of sag
P.s. I can't say it's my idea (it dates back to at least '96 as a mechanism to ensure a fixed VS/VP relationship in triode wired pentodes to avoid Is runaway)
Last edited:
I'd rather not regulate the screens because that usually obligates the regulation of the bias supply as well. A mosfet in place of the zener as a buffer for a voltage divider sounds attractive. Or, an "amplified zener" where it still functions as a series dropper might also make sense. I would like to keep things simple. If I do prefer the tone of the zener method - or my desire to for tube life wins out - I may just go buy a 50W 100v zener for $20 and be done with it. Everything except for the NTE parts seem to be unobtanium.Thanks for sharing your results, that might save some of us a lot of tinkering time in the future.
Just to close the loop on this I decided to go with a 100v series Zener (A 50W beast) and a shared 1k resistor. From idle to full output the screen voltage goes from about 230 to 145 (some from sag in the B+, some from drop across the resistor). At full output the screen dissipates about 3.5W. I tried a 500 ohm resistor (dissipation increased to 4.2W IIRC) but there was no improvement in tone so I stuck with the safer value of 1k.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.