SKYLINER build - inspired by ODS #124 and #060

[talbany]

Steven
I uploaded the pic again..Hope you can see it now

Well i am not going to question Rootz!.. Yes you are correct it all depends on where your layer of protection sits and how you run the amp..
I don't know which one Dumble used but my guess would be the same as Music man spec'd
BTW..Fender uses the R-3000 in some of their newer amps (Deville's etc) So take that for what it's worth


Tony

[Stephen1966]

Thanks, yeah I got the pic. That's the other pair of tubes on the opposite leg of the transformer... looks like he may have put a similar net on the tube to the far left, along with another of those 1 ohm monsters.

Musicman spec would be my guess too. The guys over at Fender do some cool stuff but MrD outdid them in so many ways.

[Stephen1966]

Here are the results of the first bias test. I'm calling this one, the Transformer Resistance Method. That is, after determining the resistance of each half of the OT, the voltage drop was measured from CT to plate pin, for each leg of the transformers primaries. The variac was calibrated at the beginning of the test and half way through to account for any drift in mains voltages. Wall voltage = 240VAC. The margin of error in all the measurements, was less than 5%. Tubes are 6L6GC-STRs rated at 30W (PDmax).

[Edit: the kink in the middle of the graph is about where I re-calibrated the variac]

Transformer resistance method.png

I considered using the negative bias voltage in 5V increments along the x axis but the initial tests were presenting too large an interval in the plate dissipation so I opted instead for 1mV increments of the cathode voltage from 32mV to 44mV, using the 1 ohm resistors at the test points. The bottom axis of the graph just shows the cathode bias in mV of test point, tube V7 as a baseline.

Each of the test points was measured for actual resistance and using ohms law, the current for each was averaged in the blue line.

The plates' current was averaged; only the two legs of the transformer were measured where they meet the plate pin. The two sets of data included the average Pa, for two tubes, which was then summed and averaged in the green line, representing the overall average plate current for all four tubes.

The calculated current in mA is on the left axis.

Simultaneously, plate dissipation as a percentage was calculated from the coolest (47.3%) up to around 70% (69.49%). Upon reaching 70%, the test was terminated.

The calculated plate dissipation as a percentage is on the right axis.

Several other tests are planned but surprisingly, in this one, the plate current consistently exceeded the cathode current by an average of 14%.

[Edit: this disparity could be a result of measuring the current across the OT, I would surmise that the 14% elevation of plate current above cathode current is because the OT current is the sum of the plate, and the screen currents. Nevertheless, there are a couple of other ways of measuring the plate current so I will try them and see what the findings bring.]

Next, using V7 cathode's TP mV scale as the same baseline, I will try the second transformer resistance method which is to measure the OT CT to ground and the Plate to GND for each leg of the transformer and subtract plate from CT voltages to arrive at the voltage drop that way. And again, I will measure and record all the cathode current figures for consistency. And again, it will be up to 70%, until I have more data and know exactly what I'm dealing with, I'm not going to push the tubes harder than they need to go.

Towards the end of this first test, there was a healthy amount of heat but no redplating I could see.

[talbany]

Here is what I have on the bias supply for ODSR model 100WR se# #060 Drawn/blueprinted July 1999
4X6L6GC Output tubes
Plate voltage=445VDC (Pin 3)
Screen Voltage-445VDC (470 ohm Screen res)
Biased @ -45.3 VDC Pin 5 (1k5 grid resistors)
Plate mA (per pair) per side
Side A=99.4
Side B=104.4

6L6 GC @ 70% max @ 445V would put you around 47.1 mA per tube
So given the age of the tubes in the amp this would imply he ran them very close 70% max plate dissipation @ idle for this particular ODSR

Hope this Helps

Tony

overdrive_reverb internals.jpg

[Stephen1966]

Here is what I have on the bias supply for ODSR model 100WR se# #060 Drawn/blueprinted July 1999
4X6L6GC Output tubes
Plate voltage=445VDC (Pin 3)
Screen Voltage-445VDC (470 ohm Screen res)
Biased @ -45.3 VDC Pin 5 (1k5 grid resistors)
Plate mA (per pair) per side
Side A=99.4
Side B=104.4

6L6 GC @ 70% max @ 445V would put you around 47.1 mA per tube
So given the age of the tubes in the amp this would imply he ran them very close 70% max plate dissipation @ idle for this particular ODSR

Hope this Helps

Tony

overdrive_reverb internals.jpg

Immensely valuable data. Can't thank you enough...

[norburybrook]

Here is what I have on the bias supply for ODSR model 100WR se# #060 Drawn/blueprinted July 1999
4X6L6GC Output tubes
Plate voltage=445VDC (Pin 3)
Screen Voltage-445VDC (470 ohm Screen res)
Biased @ -45.3 VDC Pin 5 (1k5 grid resistors)
Plate mA (per pair) per side
Side A=99.4
Side B=104.4

6L6 GC @ 70% max @ 445V would put you around 47.1 mA per tube
So given the age of the tubes in the amp this would imply he ran them very close 70% max plate dissipation @ idle for this particular ODSR

Hope this Helps

Tony

overdrive_reverb internals.jpg

Tony,

Thanks for this. Has anyone/you done any listening tests with differing bias? I think I've biased all my 6l6 ODS amps at 60% . Would re-biasing them hotter at 70% change the sound/feel of the amp?
If we're getting into subtleties of differing resistor/cap tone does the difference in the bias also have a sonic part to play?

M

[martin manning]

Here is what I have on the bias supply for ODSR model 100WR se# #060 Drawn/blueprinted July 1999
4X6L6GC Output tubes
Plate voltage=445VDC (Pin 3)
Screen Voltage-445VDC (470 ohm Screen res)
Biased @ -45.3 VDC Pin 5 (1k5 grid resistors)
Plate mA (per pair) per side
Side A=99.4
Side B=104.4

6L6 GC @ 70% max @ 445V would put you around 47.1 mA per tube
So given the age of the tubes in the amp this would imply he ran them very close 70% max plate dissipation @ idle for this particular ODSR

Depending upon how the current was measured, there is a range of possibilities. Since the data is give "per side," perhaps it came from the OT primary resistance method. If that is the case it is actual plate current (not cathode current) and the average Pa is 73.7% and 77.4%. And, we don't know who actually set the bias, or if it had drifted from where it was last set.

[talbany]

Depending upon how the current was measured, there is a range of possibilities. Since the data is give "per side," perhaps it came from the OT primary resistance method. If that is the case it is actual plate current (not cathode current) and the average Pa is 73.7% and 77.4%. And, we don't know who actually set the bias, or if it had drifted from where it was last set.

Thanks Martin
This is why I used the word "imply" along with "age of tubes"

So given the age of the tubes in the amp this would imply

Just presenting the evidence.
BTW.. The amp was owned by both Jackson Brown and Bonnie Raitt and the blue print was done in 1999
Tony

[Stephen1966]

There's also a short commentary about the 060 over on Rob Livesey's site. The provenance was well documented but it doesn't give technical details. You've probably all seen this already but just in case...

Hello Rob,

I had to register a new account in my wife's name because this site says my email address does not match my user name, and I find no webmaster or site contact to straighten this out, as on other sites.

I have a couple of significant Dumbles for sale. Jackson Browne's Skyline modified ODR #0060 (in Bonnie Raiit's purple suede combo cab), and el34 "Bluesmaster" Skyline #0204. The ODR was purchased from Browne by Greg Bayles at Maken' Music, then went to me in 1995. Greg ordered #0204, and sold it to me when it came in, 1997-98. It has never left my studio. I also have a Marshall model 1973 18w 2 X 12 from Brauer/Maken' set up and signed with three consecutive service dates, and signed H.A. Dumble on the chassis. Bayles and Brauer state that Eric Johnston used this one extensively on his first two solo albums in LA, out of the studio rental.

I have full provenance on all three, from manufacture on the Dumbles, and emails from Dumble on the amps.

I don't believe we have spoken, but I am aware of your reputation in the area, and thought you might like to know about the amps. There are a number of artists and producers very interested, if you think anyone else might want to take a look, please let me know.
Contact me for info and photos at thevintageidiot@charter.net , or call below, thanks.

Best regards,
Bruce Illig
Hillshire Recording Studio
Richmond Il
(815) 678 2482

https://thesubjectmatter.com/dumblearchive.html

[Stephen1966]

I'm starting to see why Aiken described the first transformer resistor biasing method "extremely accurate".

Here is the graph of my biasing attempt using the alternative method transformer resistance biasing, that is by subtracting the plate voltage from the centre tap voltage then dividing the result by the resistance. Taking into account the number of tubes and the various operations to find the plate dissipation as a percentage.

I estimated that the safe limit was 48mA putting me just under 70%.

Transformer Resistance Method (2).jpg

This graph looks a lot more complex than the last, but it includes the data from each of the tube measurements. The near straight blue, yellow, green and grey lines plot the cathode biasing figures. There's a pronounced increase in tube 9s current draw and this was the same in the last test. It might call for a bit of tube rolling to get the balance there - swap it with a tube on the higher resistance (blue) side of the transformer. Even so, these results are fairly linear. The extra line resistance of the cathode circuit is accounted for in the results and V7 was used again as the baseline up to 48 mV (roughly what tube 9 was giving me in mA).

The straight blue line at the top is the plate dissipation in %.

The two legs of the transformer which supplied the plate voltage to be subtracted from the centre tap voltage gave a very unreliable set of results though as can be seen from the roller coaster plot of the blue and brown lines. The problem, it appeared, was that with the plate and CT voltages so close, and constantly dancing around on the meter by a few tenths of a volt,introduced a much greater error prone result. In the time it took to measure one, and then the other, the voltage changed (high or low) and so at one point it appeared it dipped into a negative DC result.

Towards the end, I sharpened up my procedure by swapping the leads from the centre tap and plate in a kind of coordinated dance but I can't say this made any difference... the results are still crap.

To my mind, there are a couple of issues to consider here. When the meter is reading above 400V it can only provide a resolution of two decimal places to an accuracy of 0.1V and it appears to be rounding the results, meaning it could be off by 49mV in either direction, positive or negative. The effect of time between measurements is also present in the different CT voltages that were found at each step - 430.4/429.3; 440.03/441.05 and so on. A "dirty" mains power supply a contributing factor? Maybe. Inaccuracy in the meter? Possibly. Leads were clipped on the pins and untouched throughout so, connection? Less likely, but plausible.

Time between measurements and minute changes in voltage seem the most likely culprits behind these results. For this kind of measurement to be a success, it would take a more sophisticated set-up than a guy holding a meter swapping leads like someone bewitched. It would take a DMM that can measure high voltages to several more decimal places and a simultaneous measurement that overcame the mains voltage ripples.

These ripples might be occurring because of the reservoir capacitors I used. There is something to be said for larger mass capacitors but it wouldn't explain why the current is so steady at the cathodes by comparison.

Also, if it were a problem of the capacitors, I would have expected similar erratic voltage swings in the first test session where the voltage drop was taken off the transformer between the CT and each leg. It didn't swing there, being an instantaneous single measurement. What was puzzling about the first experiment was why the plate currents were consistently higher than the cathode currents. That, I don't know the answer to. But something about the way I measured the cathode currents?

The resistance of each cathode test point and ground was taken with the amp cold at the start and ohms law employed using the actual resistances (including that of the test leads involved). I had to include the total resistance which was higher than the 1 ohm resistance I found in the circuit factoring out the test leads resistance. Maybe that was a mistake, I would like to be corrected if so. Then, with the tiny resistors I am using (very small, 0.6W) heat from the amp may have been causing them to drift. This heat and tendency to drift may have been behind MrD's use of the big cement resistors. I haven't gone that far, ordered replacements but with high tolerance to heat and a larger mass. I'm likely to repeat some of these methods again when they are installed.

[Edit: Phil's updated link clarifies the transformer resistance method.]
https://ampgarage.com/forum/viewtopic.p ... 09#p432409

We may see some non-linear behavior creeping in because of the low resolution of the meter and voltage swings again but this has the virtue of being a single instantaneous measurement.

[martin manning]

For reference R Aiken describes and lists advantages and disadvantages of the various techniques here: https://www.aikenamps.com/the-last-word-on-biasing

I don't know why your OT primary resistance results are so scattered. It's true that measuring voltage to ground vs. across the primary will have a larger error due to round-off when measuring hundreds of volts, but assuming a four-digit DMM it's the difference between rounding to 0.1V vs. 0.001V. That looks big but, results in less than 1% difference in the calculated %Pa max. Measuring voltage across the OT primary is no more dangerous than measuring voltage to ground. The risk is shorting something with a probe tip, same as when measuring any high voltage. This risk can be minimized by clipping one probe to the CT and probing the plate leads with the other.

[Stephen1966]

... assuming a four-digit DMM it's the difference between rounding to 0.1V vs. 0.001V. That looks big but, results in less than 1% difference in the calculated %Pa max.

I'm not sure the rounding has much to do with it...

... Measuring voltage across the OT primary is no more dangerous than measuring voltage to ground. The risk is shorting something with a probe tip, same as when measuring any high voltage. This risk can be minimized by clipping one probe to the CT and probing the plate leads with the other.

Actually, I did it slightly differently and clipped two probes on the pins and measured the CT terminal with a hand held probe - even safer you might say. As I'm taking each of these readings though, I'm seeing fluctuations up to ~0.6V each time in the space of a few seconds. When the difference between CT and plate voltage is ~1V and you factor in the time it takes to move from taking one measurement and the next, I get the results as above. Still, the first method (measuring Vd across the transformer) and the cathode biasing all provide results that are practically linear at the mV resolution of the meter. Am I missing some occult technique when using/reading the meter? Hocus pocus aside, I can set it to read the max voltage and the min voltage. Should I be averaging between the high and low readings?

Edit: I will try that.

[talbany]

Here is an anode loading spreadsheet
Type the anode voltage (B+) of your amplifier into cell G4. Type the maximum anode wattage (plate
loading) into cell G5.
The Heat Meter shows the percentage of loading at idle for various bias currents in a loading range
from 10% to 120%. Most designs are biased to work in the green area.

If you notice on the heat meter has a green area (60% 70%) are within acceptable operating levels!
The question you need to ask yourself is?
At what point within this green area do you prefer to operate the given tube at the given anode voltage at various bias currents?

Do the math and use the spread sheet find the operating range and very the bias within that range to where you think the amp sounds best and try not to get too hung up on setting the bias at an exact point regardless as to what Dumble used..This is the luxury we have as builders to be able to exploit the operating range to find the exact point of operation we think works best for our personal amps. Similar to how we use say a Presence control

Tony

load.xls

https://www.duncanamps.com/software.html
https://www.duncanamps.com/technical/lvbias.html

[Stephen1966]

Here is an anode loading spreadsheet
...
load.xls
https://www.duncanamps.com/software.html
https://www.duncanamps.com/technical/lvbias.html

Great little tool which confirms my earlier calculations of the current range across the 50-60-70% range. If anyone wants to see this calculate to more decimal places, just unlock the heat meter and anode voltage cells and increase your decimal places in the number tab. Or don't there's something to be said for this being one of the easier, no-fuss "calculators" I've come across.

...
Do the math and use the spread sheet find the operating range and very the bias within that range to where you think the amp sounds best and try not to get too hung up on setting the bias at an exact point regardless as to what Dumble used..This is the luxury we have as builders to be able to exploit the operating range to find the exact point of operation we think works best for our personal amps. Similar to how we use say a Presence control
...

I'm with you on not trying to duplicate MrD in every respect. Couldn't agree more. That way madness lies! I think of the amp as just a part of the whole system, there's also the speaker, the cab, the guitar and right at the beginning of the signal chain (sorry to say) my brain. You can break his work down a million ways, study it, try to replicate it and still end up with a completely different sound... radically different. There's great value in trying to understand the design choices he made though, his design philosophy if you can call it that. I'm on my own formative journey here and at this stage I'm just plodding through the various techniques to find one I can call effective, accurate and true and not least to try to understand some of the more rarified choices he made. In the end, all the biasing will be done using the cathode bias method - the amp is designed that way - but when I find these kind of erroneous results I want to find out why. I'm just waiting on the replacement components now, then will run tests again. The definition of madness? Yeah, I'm built that way. But if I can say with x amount of cathode current, I've got y amount of plate current and z amount of plate dissipation I will have achieved my objective.

Right now, I'm planning the cab and head. I need to continue thinking about it, sure, but in the end, I need to hear it as it was designed to be heard.

[Stephen1966]

Out with the old and in with the new

I decided to swap out the heater wire... I'm still cleaning up after it and I can't say it offered any advantages. AWG 18 stranded wire (silicon) is just fine and was easier to install.

SAM_7774.JPG

Installing the flyback diodes and caps on the power tubes plates gave me the perfect excuse.

SAM_7770.JPG

I also now have 2W 1ohm resistors in there as well. It should be very robust now.