I have been using these things for plate loads in every amp that I have built since the parts first came out. I have been buying them from DigiKey. I have discovered that these little devices will oscillate in the low MHz region if given the chance. This will lead to instabillities, poor sound and in extreme cases blown parts. I found the cure to be a gate stopper resistor of 1K ohms mounted right at the "G" terminal. This works the same way that a grid stopper resistor works on a tube. I would advise the addition of this resistor to all 10M45 circuits.
The 1K stopper has shown no bad effects so far. I have used it in at least 15 amplifiers, some of which had driver response to 500KHz!
I have used these things as plate loads for the first stage in my two stage SE amplifiers. The result is more gain, less distortion, and in some cases improved high frequency response.
Many of my earlier amplifiers used the 5842 for the first stage. With a CCS load and a mosfet follower the tube sees a very high impedance load for all frequencies from DC to several hundred KHz. I get a voltage gain of 40, and a distortion below .2% out to 100 KHz. This is enough gain to drive a 300B directly.
My latest amplifier needed a voltage gain of nearly 50 in order to drive a KT-88 to clipping in triode mode. I needed to find a tube to replace the 5842, since 5842's are getting expensive. This would be a low cost amplifier, and I would not be using the mosfet follower because of its negative voltage requirement. I also wanted to run a high enough tube current through the first stage so that it would be able to drive the output stage without slew rate or bandwidth issues. I tried several different tubes with resistive loads. I could find no common tube that would give me enough gain and provide the required drive current. When I switched to a CCS load, I had no problem getting voltage gains of 50 into a 120K (grid leak resistor) load with several tubes. I settled on the often critisized 12AT7, which works quite well with a CCS load.
The voltage spec for the 10M45 is 450 volts. I have used them as plate loads for 45 tubes (as drivers for 845's). In this circuit I run a B+ of 500 volts, but the 45 will still drop at least 50 volts at maximum conduction, which puts 450 volts across the 10M45.
For high voltage applications, I have been cascoding the 10M45 with a pentode. I have one paired with a 6LW6 which allows 10 to 100 mA at 1KV. I have been contemplating pairing one with an 813 to make a CCS load for an 845, but even I am not quite that crazy yet!
IXYS makes a 900 volt version, the 10M90, but nobody has them in stock for small orders.
I have used these things as plate loads for the first stage in my two stage SE amplifiers. The result is more gain, less distortion, and in some cases improved high frequency response.
Many of my earlier amplifiers used the 5842 for the first stage. With a CCS load and a mosfet follower the tube sees a very high impedance load for all frequencies from DC to several hundred KHz. I get a voltage gain of 40, and a distortion below .2% out to 100 KHz. This is enough gain to drive a 300B directly.
My latest amplifier needed a voltage gain of nearly 50 in order to drive a KT-88 to clipping in triode mode. I needed to find a tube to replace the 5842, since 5842's are getting expensive. This would be a low cost amplifier, and I would not be using the mosfet follower because of its negative voltage requirement. I also wanted to run a high enough tube current through the first stage so that it would be able to drive the output stage without slew rate or bandwidth issues. I tried several different tubes with resistive loads. I could find no common tube that would give me enough gain and provide the required drive current. When I switched to a CCS load, I had no problem getting voltage gains of 50 into a 120K (grid leak resistor) load with several tubes. I settled on the often critisized 12AT7, which works quite well with a CCS load.
The voltage spec for the 10M45 is 450 volts. I have used them as plate loads for 45 tubes (as drivers for 845's). In this circuit I run a B+ of 500 volts, but the 45 will still drop at least 50 volts at maximum conduction, which puts 450 volts across the 10M45.
For high voltage applications, I have been cascoding the 10M45 with a pentode. I have one paired with a 6LW6 which allows 10 to 100 mA at 1KV. I have been contemplating pairing one with an 813 to make a CCS load for an 845, but even I am not quite that crazy yet!
IXYS makes a 900 volt version, the 10M90, but nobody has them in stock for small orders.
Thanks alot, that is all extreemly helpful information
I am thinking of using a CCS load using the 10m45 in my next amplifier, this amplifier is indeed two stage, but the Pre stage, and the phase inveter phase are comined, will this cause a problem if i am using a CCS load?
Also, I understand completly how to use the 10M45 as well as other CCS ICs to regulate the current, however how does the 10M45 "know" to drop the voltage to what is desired? In my case from 430v to 111v, however I am not yet sure what amount of current will be drawn
I am thinking of using a CCS load using the 10m45 in my next amplifier, this amplifier is indeed two stage, but the Pre stage, and the phase inveter phase are comined, will this cause a problem if i am using a CCS load?
Also, I understand completly how to use the 10M45 as well as other CCS ICs to regulate the current, however how does the 10M45 "know" to drop the voltage to what is desired? In my case from 430v to 111v, however I am not yet sure what amount of current will be drawn
alexmoose said:
you choose tube
then you choose cathode resistor (for biasing) ( with that you also know A to K voltage)
then you calculate resistor for 10M45
then you choose Anode supply (Ua-k plus anything to 450V decreased for half of desired voltage swing at anode)
10M45 is enough smart (as any CCS ) to eat each excesive volt your tube refuse
in fact-simple
I use a slightly different method.
1) choose the tube
2) pick the current that you want to run it at.
3) choose the current set resistor for the 10M45. I test the chips with a current meter and a power supply, since each batch seems to be different from the others.
4) choose the bias voltage needed to get a good plate voltage at your chosen tube current. Shoot for 1/3 to 2/3 of the supply voltage if this is within the tubes ratings.
5) If cathode biased, calculate a resistor value from the bias voltage and the tube current. If using LED's try to find a combination of LED's that add up to the required bias voltage.
6) Build the circuit. Measure the plate voltage. Tweak the cathode resistor ( or try different LED's ) until the plate voltage is where you want it. In practice it can be off by 50 volts or so without affecting the sound (with a good tube like the 5842). If all tubes actually matched the data sheets this step wouldn't be needed.
When the triode is being force fed a constant amount of current, the plate voltage is determined by the bias voltage. If you think about it, this is how the things amplify. The bias voltage changes with the input signal, which causes a large change in plate voltage.
Tweaking the bias voltage allows the user to find the most linear part of the tubes plate curve. If you have good distortion measuring equipment, tweak the bias to get the minimum 3rd (or higher order odd) harmonic.
If you are really dedicated you can put a pot on the 10M45, to tweak the current. Make small changes here, then play with the bias voltage. Make careful measurements and record the data. With a 5842 you will see that the sweet spot is pretty large. If you find that small changes make a big difference, pick a different tube. It will change as it ages, and no two of them will ever sound the same.
1) choose the tube
2) pick the current that you want to run it at.
3) choose the current set resistor for the 10M45. I test the chips with a current meter and a power supply, since each batch seems to be different from the others.
4) choose the bias voltage needed to get a good plate voltage at your chosen tube current. Shoot for 1/3 to 2/3 of the supply voltage if this is within the tubes ratings.
5) If cathode biased, calculate a resistor value from the bias voltage and the tube current. If using LED's try to find a combination of LED's that add up to the required bias voltage.
6) Build the circuit. Measure the plate voltage. Tweak the cathode resistor ( or try different LED's ) until the plate voltage is where you want it. In practice it can be off by 50 volts or so without affecting the sound (with a good tube like the 5842). If all tubes actually matched the data sheets this step wouldn't be needed.
When the triode is being force fed a constant amount of current, the plate voltage is determined by the bias voltage. If you think about it, this is how the things amplify. The bias voltage changes with the input signal, which causes a large change in plate voltage.
Tweaking the bias voltage allows the user to find the most linear part of the tubes plate curve. If you have good distortion measuring equipment, tweak the bias to get the minimum 3rd (or higher order odd) harmonic.
If you are really dedicated you can put a pot on the 10M45, to tweak the current. Make small changes here, then play with the bias voltage. Make careful measurements and record the data. With a 5842 you will see that the sweet spot is pretty large. If you find that small changes make a big difference, pick a different tube. It will change as it ages, and no two of them will ever sound the same.
mach1 said:
Hi Mach1,
THANKS! WOW! I like you circuit very much. I observed something particular as below which make a difference from others. This is more or less the topology that I am chasing for.
1. choke input p.s.
2. yin-yang diodes at the centre tap of the h.v. transformer to ground: what's the purpose? Will it better to use IXYS low noise Schotty diodes here?
3. ccs p.s. before the 2 vr tubes
4. the .15uf and 250R in series with the 2 vr tubes: used to correct the impedance? How to calculate their values? Need to put a resistor in parallel with the lower one to light it up first?
5. 510R cc at the gate of 10M45s: to prevent oscillation?
6. use of cc resistors: I like them too (no inductance) where precision is not needed
7. 2 cascoded 10M45s as the plate load of 12B4a: what's the approximate voltage drop here? How to calculate the value of the resistor need to set the cascoded ccs running at the desired current? Is the cascoded 10M45s much better than the single one technically and sonically?
8. optional output taken at the cathode of lower 10M45s: what's the advantage? Lower output impedance? How it is sounding compared with the normal output?
9. optional switch at the bypass of 12B4a cathode resistor
Sorry for the so many questions. Some may be stupid as I am not familar with the s.s. ones. Please feel free to correct me if I am wrong.
Thanks in advance!
Best Regards,
T.C. MA
1. 5687 (ECC99)
2. 8-9 Ma
3. according to the graph on the data sheet, it should be around 125-130ohms (I'm still in the theoretical/ developement stage of the driver stage )
4. I want to run the tube at 111v, how do I figure out the bias from this?
5. from the Bias voltage, just use ohm's law to figure out the resistor value.
I actually may be able to tweak the voltage to get the lowest 3rd order harmonic either using equipment from my old high school, or from my college next september
2. 8-9 Ma
3. according to the graph on the data sheet, it should be around 125-130ohms (I'm still in the theoretical/ developement stage of the driver stage )
4. I want to run the tube at 111v, how do I figure out the bias from this?
5. from the Bias voltage, just use ohm's law to figure out the resistor value.
I actually may be able to tweak the voltage to get the lowest 3rd order harmonic either using equipment from my old high school, or from my college next september
Snoopyma,
1. Yes, a choke input definitely sounds better in this linestage, even though there are two stage of current regulation and one voltage regulation afterwards. You only need a small cap after the choke, so use good quality here.
2. the yin/yang diodes are to provide a degree of isolation between the circuit ground and the mains earth (an earth lift) to prevent AC garbage on the mains earth affecting the sound. They should be at least 600V, and high amperage. Just use regular diodes here - there are no advantages using anything exotic or expensive. Resistors (approx 10R) are commonly used for this purpose, but I prefer the diodes.
4. Yes the 0.15uF/250R are used to flatten the impedance curve of the VR tubes. The VR tubes light up fine without a parallel resistor.
5 &6. Yes the carbon comps are used to prevent oscillation. The values are what I had to hand - anything from 470R - 1.2k should work fine They also sound very nice as cathode resistors for the 12B4 (I use 2 x 1200R 1W in parallel). Varnish or epoxy coat them to provide long term stability in a humid environment like HK.
7. In my opinion the cascode sounds better, although you will find a post on the 10M45 thread that disagrees. The difference is not huge; I suggest you try both and decide for yourself. You have to adjust the trimpots shown on the schematic to set the required currents - all 10M45 are slightly different, so I can't give you precise figures. From memory, a good starting point is 8 - 10 R for the first ccs, and 100R for the second ones. To set the first one solder in a 100R resistor before the ccs and measure the voltage drop while adjusting until it reaches 5.8 - 6V. When setting the second measure the voltage drop across the cathode resistor until it reaches 12 - 12.5 V. You must use a single 10M45 for the first ccs as a cascode arrangement will not pass enough current.
8. Yes the output taken from the 10M45 cathode provides a lower output impedance at the expense of reduced power supply isolation. It can provide a punchier sound, but may sound too agressive in some setups. Try both.
9. I put the switch on the bypass cap to determine its effect. On balance I prefer the sound without, but haven't bothered to disconnect it as I may experiment with better caps (BG perhaps) later on. Use high quality here - cerafine/ silmic at minimum.
Generally, I prefer the sound of good 5R4 and 5Y3 and GE 12B4s in this circuit.
have fun !
pm
1. Yes, a choke input definitely sounds better in this linestage, even though there are two stage of current regulation and one voltage regulation afterwards. You only need a small cap after the choke, so use good quality here.
2. the yin/yang diodes are to provide a degree of isolation between the circuit ground and the mains earth (an earth lift) to prevent AC garbage on the mains earth affecting the sound. They should be at least 600V, and high amperage. Just use regular diodes here - there are no advantages using anything exotic or expensive. Resistors (approx 10R) are commonly used for this purpose, but I prefer the diodes.
4. Yes the 0.15uF/250R are used to flatten the impedance curve of the VR tubes. The VR tubes light up fine without a parallel resistor.
5 &6. Yes the carbon comps are used to prevent oscillation. The values are what I had to hand - anything from 470R - 1.2k should work fine They also sound very nice as cathode resistors for the 12B4 (I use 2 x 1200R 1W in parallel). Varnish or epoxy coat them to provide long term stability in a humid environment like HK.
7. In my opinion the cascode sounds better, although you will find a post on the 10M45 thread that disagrees. The difference is not huge; I suggest you try both and decide for yourself. You have to adjust the trimpots shown on the schematic to set the required currents - all 10M45 are slightly different, so I can't give you precise figures. From memory, a good starting point is 8 - 10 R for the first ccs, and 100R for the second ones. To set the first one solder in a 100R resistor before the ccs and measure the voltage drop while adjusting until it reaches 5.8 - 6V. When setting the second measure the voltage drop across the cathode resistor until it reaches 12 - 12.5 V. You must use a single 10M45 for the first ccs as a cascode arrangement will not pass enough current.
8. Yes the output taken from the 10M45 cathode provides a lower output impedance at the expense of reduced power supply isolation. It can provide a punchier sound, but may sound too agressive in some setups. Try both.
9. I put the switch on the bypass cap to determine its effect. On balance I prefer the sound without, but haven't bothered to disconnect it as I may experiment with better caps (BG perhaps) later on. Use high quality here - cerafine/ silmic at minimum.
Generally, I prefer the sound of good 5R4 and 5Y3 and GE 12B4s in this circuit.
have fun !
pm
Look again, I just put it here:
http://www.tubelab.com/CCS circuits.htm
It was done with an output stage in mind, but the circuit will work with smaller tubes.
http://www.tubelab.com/Active loaded SE output.htm
mach1 said:
Hi mach1,
Thanks really for your informative reply. I will definitely try the cascoded 10M45s. But what's the maximum current that they can supply?
One more question: how to calculate the values of capacitor/resistor which is used to flatten the impedance of the series vr tubes?
Thanks again.
Regards,
T.C. MA
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.