Hello, I'm constructing a 300B Single Ended Headphone AMP for my friend. And I'v finished the design and test of a Filament Constant Current Source for DHTs, using the schematic based on the public version of Rod Coleman's Kits.
But now I have a severe problem that my friend has a pair of Takatsuki 300Bs on his present AMP, and it's hard for him to test the actual filament current by himself. My filament kits can only provide current higher than 1.3A due to the resistor values, should I change the resistor values for a lower minimum current?
Any actual filament current about 300Bs provided will be of great help, Thanks🙂
(Actual means accurate 5Volts on the pins of the tube)
Here are some pictures😏
But now I have a severe problem that my friend has a pair of Takatsuki 300Bs on his present AMP, and it's hard for him to test the actual filament current by himself. My filament kits can only provide current higher than 1.3A due to the resistor values, should I change the resistor values for a lower minimum current?
Any actual filament current about 300Bs provided will be of great help, Thanks🙂
(Actual means accurate 5Volts on the pins of the tube)
Here are some pictures😏
The data sheet says 1.2A. This is a voltage drive filament, so when you want to drive it with current you can do several things. Increase the current until the voltage is 5V, or set the current at 1.2A.
Or drive the tube the way the designers intended, with 5V
Jan
Or drive the tube the way the designers intended, with 5V
Jan
Yeah, It's a voltage drive filament, but I've already get a positive feedback from my friend about the sound using this CCS kits. And my design focus on simplify the signal path from all possible aspects. The filament CCS is a vital step blocking unnecessary signal paths, so I don't want to give it up.
I'v asked SuperTNT about this issue. And they replied me that Chinese Direct Heated tubes always required higher current than the datasheet says.
The current value listed is only there for design reference, such as the power capabilities of the filament transformer windings.
If I set the current at 1.2A for my 300BF, the voltage applied on the filament is only 3.85 Volts, which is totally harmful for the tube using as a power stage.
So, what I want to know is the current values when drive it at 5 Volts DC.
Thanks for your reply
SuperTNT provided me the AC filament current test data of the four tubes. They are almost the same, as well as the curves. So the current of one can represent all of the set.
The key problem is that I have no idea about the current of tubes from diffenent factories，since the filament material and the manufacture techniques may be different.
Ordering tubes of other brands is obviously not a good choice for me😵
On the site of Takatsuki it says that the filament current is 1.2 A:
Takatsuki 300B
These tubes sell for around $ 2,500.00 per one so I would think that a little aftersales is in order. Why don't you email Takatsuki to ask if the stated 1.2 A is trustworthy?
Takatsuki 300B
These tubes sell for around $ 2,500.00 per one so I would think that a little aftersales is in order. Why don't you email Takatsuki to ask if the stated 1.2 A is trustworthy?
Make the CCS adjustable, and tell the user to adjust the current for 5V across the heater.
Works for any tube.
Of course you will need to re-adjust over time as the tube ages, but that will only add to the magic.
Jan
Works for any tube.
Of course you will need to re-adjust over time as the tube ages, but that will only add to the magic.
Jan
Actually All 300Bs says the current of their filament is 1.2A, except those enlarged version such as VV52B.
It's adjustable, but with the present resistor values, the minimum output current is 1.25A.
The actual current of an individual 300B is never specified. Instead, The voltage must be set to 5.0V, and the current will be what it may. The data-sheet value for current is only a approximate guide. It is usually within 10% of the data-sheet guide-value, but sample-to-sample variations are too large to be able to set a fixed current in advance, and expect to give the correct Vf with random 300Bs. If you do that, the tube will have a short life, and possibly a very short life.
Current-driven regulation is usually used, because voltage regulation of the filament sounds so bad (as you would expect, after analysing the reaction: the anode-current (the music signal) gets diverted around the regulator, as the feedback tries to correct for it). Current-driven heating gets around this, but it must be adjustable with a trimmer, to get around the sample-to-sample variations. Power up, adjust current until 5.0V appears across the filament, and monitor for a while to see that it is stable.
With a 300B, the filament current might be anywhere from 1.1 to 1.4A. Plugging a 1.1A (@ 5V) 300B into a socket that delivers a fixed 1.3 or 1.4A will destroy the coating of the filament in a very short time, as the voltage will not just be (1.4/1.1)*5V, but higher still: the resistance is temperature-dependent. Please don't do this to a lovely Takatsuki 300B!
The circuit you're working from is a very old suggestion (almost 20 years old). It has no way to adjust it. It is not temperature compensated, so the output will start high, and drop by ~10% as the circuit warms up. This will also affect the tube life. It was never built with the transistors you are working with, and those MJL42xx series parts have very high capacitance that may make the circuit unstable without re-designed frequency compensation. The new V9 fixes all these shortcomings, and has much better ways to minimize unwanted feedback into the regulator, that have a notable effect on the sound quality.
Current-driven regulation is usually used, because voltage regulation of the filament sounds so bad (as you would expect, after analysing the reaction: the anode-current (the music signal) gets diverted around the regulator, as the feedback tries to correct for it). Current-driven heating gets around this, but it must be adjustable with a trimmer, to get around the sample-to-sample variations. Power up, adjust current until 5.0V appears across the filament, and monitor for a while to see that it is stable.
With a 300B, the filament current might be anywhere from 1.1 to 1.4A. Plugging a 1.1A (@ 5V) 300B into a socket that delivers a fixed 1.3 or 1.4A will destroy the coating of the filament in a very short time, as the voltage will not just be (1.4/1.1)*5V, but higher still: the resistance is temperature-dependent. Please don't do this to a lovely Takatsuki 300B!
The circuit you're working from is a very old suggestion (almost 20 years old). It has no way to adjust it. It is not temperature compensated, so the output will start high, and drop by ~10% as the circuit warms up. This will also affect the tube life. It was never built with the transistors you are working with, and those MJL42xx series parts have very high capacitance that may make the circuit unstable without re-designed frequency compensation. The new V9 fixes all these shortcomings, and has much better ways to minimize unwanted feedback into the regulator, that have a notable effect on the sound quality.
Thanks for your attention.
I think I'm clear about why the CCS filament sounds good. Just like your discription, it blocked unnecessary signal paths.
I know the virations are unguaranteed, so what I'm doing is to get some samples to set a proper adjustable range.
The viration of filament current may be the biggest obstacle for the CCS filament to be applied in mass produced amplifiers. I'll test the current for every tube and adjust the current to a proper value before install.
I've made a lot of changes from the public version, such as adding a CRD together with a resistor in the Gyrator section to set an adequate Vce for the darlinton transistors. Adjustable is not a problem. It seems none has seen the trimmer in my pictures.😵 Actually the final circuit is really similar to your V8 version.😏
The temperature coefficient is what I'm concerning at first. My first CCS kits prototype was build with Kelvin connection 4 wire shunt resistor and OPA187 which has a perfect Vos tc together with net resistors. But finally, it goes into oscillation and I gave up using OPAMPs. (Heating the OPAMP and resistors with 100℃ air flow only caused about 1mA changes in the current, but it's always oscillating.) Finally in my version I placed an NTC near the reference BJT and culculated the Vbe temperature coefficient carefully. I certainly noticed that.🙂
I use the MJL4281 on this mainly because I set a much higher and more stable voltage drop on the BJTs. For example 2Volts on the resistor and 1.5Volts on the Vbe of darlinton transistors finally created a voltage drop of 3.5Volts in total on the power transistor, and will create a huge cooling problem. According to simulations, my Gyrator section can provide over 100Ohms of AC impedance from 100Hz to 20KHz.
I helped my friend to assemble a pair of your V8 version for 2A3s, from which I could see that you're not using the Gyrator in the upper section, Why? (I've found that changing the Gyrator into a rough voltage regulator could be helpful for the stability in my OPAMP prototype. Is that the reason?)
Stability is not a problem for me because I can catch any ripples immediately. The background noise of my oscilloscope is below 30μVrms in 50Ω high resolution mode , and below 100μVrms in 1MΩ directly sample full bandwidth mode. At the top of my PCB, there's a place for a MCX socket with a 4.7μF MLCC capacitor for AC coupling, which allows me to connected the output directly to the oscilloscope to check the noise in 50Ω mode with least disturbance.
Actually I've received complaints on yours about poor material selection. That's also a reason why the material selection of mine is a little bit exaggerated. I'll keep seeking better parts.😎
Last edited: