I'm a longtime audiophile and over the last 18 months have been doing more DIY projects. I now have a half dozen projects under my belt (no tube designs except for a tube input buffer on a class D amp) and I'm thinking my next project should be a pair of 300B SET monoblocks.
My current speakers are 93db/w, but I'm working on a new pair that will be 99db/w line sources with a nice easy load impedance. These will cross over to separate active subs at around 180hz, so these amps won't have to handle the lower few octaves. So I'm figuring something in the 4-8w range will be plenty.
I'm not trying to save money by building my own - I do DIY for the enjoyment and satisfaction - so I'm looking for an amp project that will deliver stellar performance without much concern for build cost.
I'm planning to use a balanced connection, so will either need a balanced input or can add a coupling transformer.
I'm willing to consider a kit if the parts quality is high, but I'm comfortable building from scratch.
I spent some time searching this subject but the threads I found with recommendations were pretty old with mostly broken links. So I'm looking for recommendations on where to start.
Thanks,
Jay
My current speakers are 93db/w, but I'm working on a new pair that will be 99db/w line sources with a nice easy load impedance. These will cross over to separate active subs at around 180hz, so these amps won't have to handle the lower few octaves. So I'm figuring something in the 4-8w range will be plenty.
I'm not trying to save money by building my own - I do DIY for the enjoyment and satisfaction - so I'm looking for an amp project that will deliver stellar performance without much concern for build cost.
I'm planning to use a balanced connection, so will either need a balanced input or can add a coupling transformer.
I'm willing to consider a kit if the parts quality is high, but I'm comfortable building from scratch.
I spent some time searching this subject but the threads I found with recommendations were pretty old with mostly broken links. So I'm looking for recommendations on where to start.
Thanks,
Jay
If you want a little "exotic" (and not disturbed by higher voltage and higher impedance output transformer) try any thoriated tungsten filament tube (for example NOS 10/VT25, 801/VT-62 or new Elrog tubes).
See Vinylsavor site: VinylSavor
See Vinylsavor site: VinylSavor
Last edited:
I've spent the last year building, rebuilding and listening to several versions of a 300b amp. I found out a lot on the way. I ended up with the simplest and most predictable outcome - a 2 stage design using a triode input tube with a plate choke and no cathode bypass cap and a teflon coupling cap (FT-2). I wanted a DHT driver like 10Y, but since there aren't any really good ones with enough gain, I settled for a good sounding triode. I eliminated a DHT with a step-up input transformer because the transformer degraded the sound, and the same to a lesser extent with an op-amp on the input to increase the gain.
I've been doing cathode bypass tests and on the 300b in self bias by far the best are the DC Link range of caps. They were a bit of a revelation and for a 300b in self bias, these are indispensable. But better still is no cap. So on the input tube I used a plate choke with inductance of around 150H which allowed me to use an unbypassed cathode resistor around 270R. Lundahl do good plate chokes, but so do others. Mine are amorphous core from NP Acoustics, Vietnam.
Choice of input tube will be a triode, or pentode in triode, of mu=30 or a bit more. Popular choices range from E80CC to C3g and D3a. Others are E180F, E810F, E280F, EC8010 with care to avoid oscillations with ferrite beads etc etc.
For the output transformer I preferred the sound of a 5K primary - cleaner and just better. I had a Lundahl LL1664/70mA to use but a LL1682/50mA simply sounded better. I expect a LL1663/50mA would be good too. I used a relatively low operating point around 270v, so more in 45 territory. Rod Coleman regs for the DC filaments.
I don't use a balanced system. If you need an input transformer to convert balanced to SE this will be a potentially weak spot and I'd recommend an amorphous core. Better avoided altogether if you have an alternative arrangement.
.
I've been doing cathode bypass tests and on the 300b in self bias by far the best are the DC Link range of caps. They were a bit of a revelation and for a 300b in self bias, these are indispensable. But better still is no cap. So on the input tube I used a plate choke with inductance of around 150H which allowed me to use an unbypassed cathode resistor around 270R. Lundahl do good plate chokes, but so do others. Mine are amorphous core from NP Acoustics, Vietnam.
Choice of input tube will be a triode, or pentode in triode, of mu=30 or a bit more. Popular choices range from E80CC to C3g and D3a. Others are E180F, E810F, E280F, EC8010 with care to avoid oscillations with ferrite beads etc etc.
For the output transformer I preferred the sound of a 5K primary - cleaner and just better. I had a Lundahl LL1664/70mA to use but a LL1682/50mA simply sounded better. I expect a LL1663/50mA would be good too. I used a relatively low operating point around 270v, so more in 45 territory. Rod Coleman regs for the DC filaments.
I don't use a balanced system. If you need an input transformer to convert balanced to SE this will be a potentially weak spot and I'd recommend an amorphous core. Better avoided altogether if you have an alternative arrangement.
.
Last edited:
Please tell me which company does layer wound plate chokes for the price and quality of Lundahl? What can be bought on every corner of the internet is random wound choke coils.
This is for some reason.
You can buy fancy cores with random wiring in Vietnam. But I don't expect them to be layer wound.
So never compare apples with pears. The quality of an audio choke is determined by core material AND winding style.
Last edited:
Speakers are single unit Alpair 10M, gen. 1. Listed as 7.2 ohms. The LL1682 is 5K into 5 ohms, so a bit more into 7.2 ohms even. With these units the LL1682/50mA certainly sounded better than the LL1664/70mA. No contest. The 50mA rating would mean higher induction as well, so this would be a factor.
Markaudio Alpair-10 Gold 6" Full Range Gen 2
5-8W are perfectly fine with the 93 dB/1W speakers too. Plenty of choice for SE amp, assuming the load your speakers provide is suitable: 45, 2A3, 10Y/801A, just to name the most common among DIYers. Parallel SE included.
I am not a great fun of 300B but for 8W max, the input/driver stage can be reasonably straightforward.
Thank you Andy, that is certainly a nice speaker!
I see in the datasheet that it has an impedance going from 7.5 ohm at 300Hz to ~17 ohm at 20kHz.
The LL1682 reflects the secondary load (+ the secondary DC resistance of 0.4 ohm) 1024 to 1. So your primary load is between 8K and 18Kohm, depending on the frequency - quite a lot for a 300B.
This means:
- the output power is reduced
- the THD is also reduced
- the OPT is underdamped, so its high frequency response is peaking. This may add "sparkle" to the treble. It shows as ringing on square waves on the oscilloscope.
- the bass response is reduced.
It doesn't sound too good on paper, but I agree that it can sound good in practice, depending on the speaker.
I see in the datasheet that it has an impedance going from 7.5 ohm at 300Hz to ~17 ohm at 20kHz.
The LL1682 reflects the secondary load (+ the secondary DC resistance of 0.4 ohm) 1024 to 1. So your primary load is between 8K and 18Kohm, depending on the frequency - quite a lot for a 300B.
This means:
- the output power is reduced
- the THD is also reduced
- the OPT is underdamped, so its high frequency response is peaking. This may add "sparkle" to the treble. It shows as ringing on square waves on the oscilloscope.
- the bass response is reduced.
It doesn't sound too good on paper, but I agree that it can sound good in practice, depending on the speaker.
Yes - on paper is shouldn't work, but I love the sound so in practice it works just fine. Frequency response seems OK to me, though I haven't measured it. It goes loud enough for me even at relatively low HT voltage. I'm very pleased with this OPT.
I trust your ears. This shows that experimenting with unusual configurations can give good results.
Going back to the original topic - I vote for Tubelab SE II, it's a good design.
What does the impedance actually look like. Often, with a typically high Rout SET, this is more important than sensitivity.
dave
thread open.
Nope. Lets assume first that the load is resistive. The high frequency behaviour does not depend on the reflected primary impedance. It depends on the source impedance (300B plate resistance plus DC series resistance), leakage inductance and parasitic capacitance. All these do not change if you change the load. Then let's take into account inductive behaviour of the speaker. This is pretty common behaviour, even for tweeters, so I would not expect anything strange. The inductive rise of impedance actually works as an equalizer, being SE amp with low DF, and this might compensate for the lower power available.
The low frequency cut-off (-3 dB) is f= Req/2*pi*L. Req is the the equivalent resistance give by parallel of the 300B plate resistance and the primary load.
Assuming that the 300B has 750R at 50 mA:
1) If the primary load is 8K then Req = 686R
2) If the primary load is 18K then Req =720R
3) If the primary load is 5K then Req = 652R
There is very little difference....thanks to the low plate resistance of the 300B, of course.
Almost correct.
In the equation of Req you forget series resistance of the primary winding.
Let's assume primary DCR of 100 ohm, typical of a good quality 300B output transformer, then your values should be 768R, 811R and 726R respectively.
Not that it matters that much, but when you demonstrate with a calculation example, take primary DCR into account.
It is an important and quite often overlooked parameter and can be the difference between a so-so transformer and a really good one.
For Andy's LL1682 Req is in the order of 817R.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.