• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

300B SET amp measurements without tubes installed

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Is it ok to have power on a SET amp without tubes inserted?
Is it also ok in a 300B SET direct coupled amp with ECC83 as the driver only to have the ECC83 or only the 300B inserted?
I want to make some tests to try to find out where hum is comming from. E.g. if there is hum on the output transformer without any tubes inserted this will tell something etc........
 
We need a schematic to more fully answer as to how to troubleshoot the hum.
Not just simply a schematic, but with component values too.

The 300B will not get its proper bias with the ECC83 removed (direct coupled, right?). It may blow the 300B, the output transformer, etc.

Can your filter caps stand the higher B+ voltage when the 300B and ECC83 are removed?

By the way, did you use a bleeder resistor on your B+ that will discharge the caps when you power down, and go in and get your hands in there?
Safety first!
And always use a DMM to check that the B+ is discharged before working on the amp.

If you remove the 300B, the hum could be caused by either 'the power transformer or the power supply choke (assuming you still load the power supply)', which then couples magnetically to the output transformer.
Are the output transformer windings oriented 90 degrees versus the power transformer and choke windings?
How far is the output transformer to the power transformer and the choke?

Did you use AC or DC filaments on the 300B?
If it is AC, did you use a hum balance pot?
Even with that, some hum may be present.
It depends on the exact 300B, some are worse than others, even from the same manufacturer.
I can give an explanation for that (mostly unknown effect) if that is the problem (ask
me).

How much filtering is in the B+ power supply, for the 300B and for the ECC83?

By the way, direct coupled is much harder to troubleshoot hum, because you can not just short the 300B grid to ground (you can for most RC coupled and most interstage designs).
Which brings us back to the need we have to see your schematic.
I have some tricks that might work, depending on your schematic.

A picture of the top of the amp and transformers and choke might help too, but not if they are potted
(can't see the winding orientations).
 
Last edited:
Thank you for the answer. I have some hand drawn schematics of each tube (made yesterday and today). I have both 50 Hz hum and also some higher frequencies (probably from PSU). The power transformer is oriented 90 degree to the output transformers. There is a star ground. 300B uses 5V DC (a hum pot is there anyway....or whatever the purpose is). The 6.3 V AC for the ECC83 has two 56R to ground. The PSU is not filtered very good. From the bridge a 220 uF cap and from there a 220R filter resistor and then again a 220 uF cap. This is all. B+ is approx. 535 V and caps 550/595. It is direct coupled......no caps in the signal path. Hum level at the 8 ohm ouput is about 4 mV for the one channel and 2.5 mV for the other. I can hear the 50 Hz on my 95 dB speakers. Not loud.....but still. Even with hum they play amazing.

Does this give more information?
I have re-wired the 6.3 V AC for the ECC83 a little bit. There was a loop around the sockets. Now it goes straight over.
There is a loop also on the 300B socket for the DC.....but I assume this is ok.

A picture from the bottom attached.......then the *-ground etc. is visible.

Updated schematics for the ECC83.....found some errors on component values......
 

Attachments

  • DSC_3669.jpg
    DSC_3669.jpg
    361.1 KB · Views: 578
  • DSC_3671.jpg
    DSC_3671.jpg
    136.2 KB · Views: 566
  • DSC_3654.jpg
    DSC_3654.jpg
    503.8 KB · Views: 559
  • DSC_0843.jpg
    DSC_0843.jpg
    442 KB · Views: 531
  • DSC_0837.jpg
    DSC_0837.jpg
    937.1 KB · Views: 540
Your schematic makes it look like there are some mechanisms to get power supply noise injected into the amplifier to quiet it down. There is also some global feedback which should quiet the noise down a bit.

Have you tried adjusting the hum pot?

4mV and 2.5mV seem very high for a DC heated 300B amp. This is especially true since you have a 7805 regulator on each tube; this should be very quiet. For comparison, the 300B stereo amp I use has regulated DC on its filaments but a bit more power supply filtering, and I see something like 0.6mV of hum on the outputs.

You can buy a pair of 4K/50W resistors and solder one from pin 2 on each 4 pin socket to the wiper of the hum pot, lift each feedback resistor, then run the amp. This will let you listen to the power supply noise component of what you're hearing.

You can put your scope on the output of the filament supply to see how quiet it is (with a 300B connected so it's loaded).

Have you contacted the manufacturer?
 
I have tried to adjust the pots but it has no real influence on the hum. And that was expected when it is driven by DC? .....what I was able to was to adjust the cathode voltage a little bit (2 volts) so each tube is approx. identical (180 V which gives 60 mA of bias current).
The idea with the 4k/50 W resistor was instead of having the 300b tube installed?
Sounds like a good idea.
I have talked with the manufacturer and it is possible to get some help. I have also contacted a small company from the Netherland called "Tentlabs" and it seems I can get some good advices from there and make it as a small DIY project.....instead of shipping the amp back. Then I will learn something about tubes :)
The "Tentlab" person seems to know a lot about tubes, noise etc. both theoretical and practical. I am most concerned about the 50 Hz hum......so I am interested in knowing the source for this 50 Hz......
 
Yes, the hum pot with DC can be a bit overkill, especially since you have a few DB of feedback.

Tent knows what's up, I wouldn't hesitate to take a drive and visit him if he's up for it.

50Hz is your mains frequency. In an amp like yours, we would expect you to get 50Hz noise from AC heating of the 300Bs. Your power supply noise will be 100Hz, so one other place you might look is the driver tube heaters.

If you short the 47K resistors at the inputs, do things change?
 
That was the next step I would try.....to short the 100k (first diagram was wrong) resistor at the input.
I did it at the RCA inputs. Doing this I got these results:
AC left channel: 4.2 mA to 3.7 mA (has the longest shielded wire to the input tube)
AC right channel: 2.6 mA to 2.4 mA.
I made the measurements several times with and with out input grounded so I am sure that there was a significant change.
 
Some additional measurements (measued by using a multimeter only):
Short the input at the ECC83 input did not make any "hum change" compared to short the input at the RCA connector.


Measured DC on 300b heater to be 4.95V.
Measured AC on the 5V DC on the 300b to be 55 mV <-- can that cause hum?
Measured AC on the B+ to 8 V <-- is that much?


Could be interesting to look at is using a scope.
 
I. 8V ripple on 300B B+ is far too much.

Example: 300B rp = 700 Ohm, OPT primary = 3000 Ohm, secondary = 8 Ohms.

Root (8/3000) = 1/19.4 the voltage ratio of the output transformer.
8V * (3000/(3000 + 700)) is the ripple across the 3000 Ohm primary
8V * 0.81 = 6.48V across the 3000 Ohm primary
6.48V/19.4 = 0.334V (334mV ripple)

II. 300B u = 3.85
The gain of the 300B and output transformer is 3.85 * (1/19.4) = 0.198
That is the gain from the 300B grid (relative to the 300B filament), versus the output at the 8 Ohm tap.

Now, suppose you have 55mV ripple on the 5V filament, and you do not either use a balance pot, nor at least use some matched resistors on the filament to connect their center to the bias resistor. There is no balancing of the 55mV ripple.
In that case, you would connect one end of the filament (+ or -) to the bias resistor to ground.
Then:
55mV * 0.198 = 11 mV ripple at the 8 Ohm tap.

Is the hum mostly 50Hz, 100Hz, or both?
That is also a clue to where the hum(s) is/are coming from.
Full wave rectification should have dominant 2x line frequency (i.e. 100Hz).
Half wave rectification should have dominant 1X line frequency (i.e.50Hz).

Ironically, AC filaments on a 300B will usually have dominant 2X line frequency.

Hint: Never rely on negative feedback to reduce hum, get the hum reduced first, and then the negative feedback will not have to "work so hard".
And by the way, the "hum error correction signal" that is fed back will go back through the amplifier, and will be distorted.
50 Hz hum will now have 100 Hz and 150 Hz 2nd and 3rd harmonic distortion.
Yes, those 100Hz and 150Hz distortions will be partially corrected by the 2nd time around of the negative feedback correction signal, but 2nd and 3rd harmonic distortion will now make them 100x2, 100x3, 150x2 and 150x3Hz.

For more on how negative feedback creates upper order harmonic distortions, look up some Norman Crowhurst documents from the 1950s.

I generally shoot for less than 500uV ripple on single ended amps, and less than 300uV ripple on push pull amps (and they do not have any global negative feedback).

I hope that helps.
 
Last edited:
Thank you very much for the answer and calculations.
I need to look at the measuremets using a scope to be sure my measurements are correct. Also to check how the waveforms look like. To my ear there was some 50 Hz from the woofer. Also some "buzz" from the mid speaker.
The NFB was put there (by the maker) to give the amp a more solid bass response. I was told I cold just remove it if bass response was "too much". It seems a good idea to remove it.......at least until the hum problem is solved.
I have plans for a 10H choke just in front of the diode bridge. This should reduce the ripple? .....the one I am looking at has a resistance of 130 ohm as far as I remember so I will loose a few volts......but I guess it should be ok. It can be placed in front of the power trafo. Should it be turned 90 degree to the power trafo?
I noticed that the wire bundle you see on the picture consist of the 400V AC, 6.3 V AC, 2 x 5V DC. Can that cause a problem? ....can that cause some noise to be inducted into the 5V DC? the 400V AC is going direct to the bridge and after that a fast 220uF Jensen audio grade capacitor. There will be some current peaks in the 400V AC wire which could be transferred to the 5V DC wires?

By the way....I was not able to reduce the 55 - 65 mV ripple at the 5V DC by trimming the hum pot.
The amp as a soft start so after approx. 45 sec the B+ is put on. I will measure how the ripple at the 5V DC looks like before and after the 45 sec. It was yesterday I discovered how the softstart was implemented and that I can use it in the measurements to check where the hum is generated.
 
The pot does not reduce the 55mV ripple. It merely applies a different amount/portion of the ripple to the 2 ends of the filament.
Suppose one end of the filament versus the grid voltage is more sensitive than the other end of the filament versus the grid voltage. i.e. 40mv to the less sensitive end of the filament, and 15mV to the more sensitive end of the filament (and so tends to cancel the effect, because it is the opposite phase).

Bundling the HV wires with other wires is probably not the best construction technique.
I would worry more about the 400V and 6.3V wires bundled together. The high gain ECC83 / 12AX7 might have some capacitive 'leakage' from the filament to the cathodes. That might be the cause of the higher frequency buzz in the midrange (capacitive 'leakage' amplitude increases as the frequency increases)

For a power supply, the order is transformer secondary, rectifier(s), choke.
It is not transformer secondary, choke, rectifier(s).
 
Some additional measurements (measued by using a multimeter only):
Short the input at the ECC83 input did not make any "hum change" compared to short the input at the RCA connector.


Measured DC on 300b heater to be 4.95V.
Measured AC on the 5V DC on the 300b to be 55 mV <-- can that cause hum?
Measured AC on the B+ to 8 V <-- is that much?


Could be interesting to look at is using a scope.

Firstly, be careful about connecting a scope to the filament. Please remember, the scope ground should be connected to safety ground, and the amplifier ground/chassis is (should) also be connected to safety ground. If you connect the probe-ground to one end of the filament, you will short-circuit the cathode resistor network, and the anode-current of the 300B will increase dramatically, and risk damaging the 300B.

You can safely measure the filament noise with the anode voltage switched OFF.

But in any case, 55mV at the output of a 7805 is not a good result, and with 95dB/W speakers, you will have difficulty getting it quiet.

The problem is likely to be that the input voltage to the 7805 is too low to achieve good supply-rejection (it needs over 8V when the mains supply is at its lowest), but it must also support the range of output current required by different 300Bs (1.2 - 1.4A).

So, what voltage do you have at the input of the 7805?

I see that the 2 7805 regulators are mounted close together on the same heatsink. Increasing the supply voltage will risk overheating the 7805, especially when the mains-supply is at high tolerance. The older Audio Note Kit-1 had problems like this. The The 7805 has a thermal protection for itself (it cuts out when near the thermal limit), but cutting the heating power while the anode voltage is still present is not a good idea.

Reducing the ripple at the input to the 7805s can reduce the noise, but bigger supply-capacitors means higher rms current through the transformer and more stress & heat on the rectifiers - so it should only be done if all the design considerations are understood.
 
The 55 mV AC on the 300b filament is there without B+ is on. So it must be the 5V DC which is "noisy" by itself?
I measure 7.7V AC as input to the rectifier bridge and 8.5V DC on +- terminals of the rectifier bridge. 7805 gets what it needs to be able to regulate properly?
About my scope and measurement on the 5V filament. My scope is a LabNation SmartScope and connects to an Android tablet so no common safety ground (battery driven).
I see the problem using a real scope if the probe ground is the same as the amplifier ground......
I should be able to use the SmartScope as I use my multimeter.....i assume/hope....
Could a capacitor direct on the 300B filament reduce the AC noise?
What size would you recommend?
Something that has been done before?
About extra B+ filtering. In the moment it is as follows:
secondary : Rectifier bridge : 220uF capacitor : 220 ohm filter resistor : 220uF capacitor.

The idea was to insert a 10H (130 ohm) choke just after the rectifier bridge and keep the other as is. Would that work? .....how much less ripple can I expect?

About the wiring I have attached a picture of the big bundle of wires:
Twisted brown: 400V AC (B+)
Twisted white: 6.3V AC (ECC83 heating)
Twisted gray: 6.3V AC back from ECC83 and used to the 555 timer board for delayed B+ (45 sec).
Twisted white/red
Twisted white/yellow: 5V DC to 300B filament
Black: ground
Green: NFB wire back from OT to ECC83 for left channel (the one which has most hum).

Do you think I should spilt bundle and rewire some of it?
 

Attachments

  • DSC_3672.jpg
    DSC_3672.jpg
    358.6 KB · Views: 228
Yes, the 55mV in the dc filament supply output is caused by the ripple voltage on the chip's input. The capacitors are probably too small (they look too small) and so the ripple voltage will be too large, and the voltage will drop below the minimum value to maintain low noise. Electrolytic Capacitors that are too small will also have a short life, because they can't withstand the ripple current (the rms value of the pulsating current from the rectifiers: nearly 2.3A for a 1.2A dc rectified supply).

low dropout regulators like the MIC29300 that Mel suggests can reduce noise, but beware that they are not always a drop-in replacement: to get low dropout they use PNP output transistors, which greatly increase the open-loop output impedance. This means that the caps at the input and output must be as per the data sheet: No ceramics or other low impedance film caps, but bigger electrolytic values than for 7805. These old 1990s chips will oscillate if the caps are outside the specified range of C-value and ESR. For all the work involved in swapping the chips & fixing these capacitor problems, you still have a voltage regulated supply, which can't distinguish between controlling the filament heating (good), and destroying the differential part of the music signal, across the filament (bad). A current-driven filament supply avoids this flaw.

10H choke for extra B+ filtering? Yes, if it is rated for enough current (say 120mA for 1 300B, or 240mA for stereo pair) then it will improve the B+ noise, yes.

Search for the "Duncan PSUD2" application, to model the B+ supply, and see how much improvement in ripple you get.
 
About the wiring I have attached a picture of the big bundle of wires: <snip>

400V ac from the transformer to the rectifier, and the wiring from the rectifier to the B+ capacitor should be kept short. Very short - and kept right away from any signal wiring (the 300B filament wiring should be treated as signal). The voltage swing, and the current pulses in these wires mean that they will emit large electromagnetic fields. Keeping them short minimises the problem - having them long and close to other wires makes the problem much worse.

The 300B filament wiring should certainly be kept separate from all other wires.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.