I'm calling this Version 1.1.
I'mm keeping the output bias resistors at 100Ω, as 56Ω runs things way too hot.
An OPA1652 works better than the OPA2132 for U3 (and is cheaper as a bonus). LF output offset now under 20mV.
OPA1662 & OPA1612 opamps otherwise reduces the noise compared to the NE5532. I also increased the gain of the first stage to capitalise on this. I'm measuring 300nV/√Hz at the LF output at 1KHz with 33dB (x44) overall gain - that's 6.8nV/√Hz input referred, so reasonable given all those opamps.
I reduced R68 (shunt resistor on tweeter power amp) from 1K to 470Ω. This knocks the output offset for the tweeter stage down under 20mV to match that of the woofer.
I've also ditched the 2N5087/89 in favour of 2N3904/06. I see no improvement in practice using the 2N50XX parts.
Gotta say I'm really happy with this amp - it's the quietest amp I've done so far - absolutely no hint of any hum or noise at all - even driving the output into headphones. Plus it's got oodles of power to run the noiseUnit drivers to the point where the voice coils are hitting their stops.
I'll be making a bunch, I reckon.
I'mm keeping the output bias resistors at 100Ω, as 56Ω runs things way too hot.
An OPA1652 works better than the OPA2132 for U3 (and is cheaper as a bonus). LF output offset now under 20mV.
OPA1662 & OPA1612 opamps otherwise reduces the noise compared to the NE5532. I also increased the gain of the first stage to capitalise on this. I'm measuring 300nV/√Hz at the LF output at 1KHz with 33dB (x44) overall gain - that's 6.8nV/√Hz input referred, so reasonable given all those opamps.
I reduced R68 (shunt resistor on tweeter power amp) from 1K to 470Ω. This knocks the output offset for the tweeter stage down under 20mV to match that of the woofer.
I've also ditched the 2N5087/89 in favour of 2N3904/06. I see no improvement in practice using the 2N50XX parts.
Gotta say I'm really happy with this amp - it's the quietest amp I've done so far - absolutely no hint of any hum or noise at all - even driving the output into headphones. Plus it's got oodles of power to run the noiseUnit drivers to the point where the voice coils are hitting their stops.
I'll be making a bunch, I reckon.
Attachments
Hi Suzy,
PCBs version 1.1 are on their way from China to Germany. Already ordered most parts for the Noiseunit, maybe i try the BC546/556 B Version instead of 2N3904/3906, with a Vceo of 65V. I have seen them often as differential input stage in poweramps. I will give them a try.
I have decided to go for 15V AC, since you pointed out that this design is not scalable to any AC voltage in the range of 12V to 18V AC. You also mentioned, that there are more effektive designs for the power class of 25W /8Ohms.
I think, that heatsinking at 15 V DC is still managable with your design. The availbale output power will be enough for my small office.
I found this site 404 Not Found for recalculation of the filter stages. So i can adopt the active filter to my needs.
Did you make any further changes in the design?
BR to DownUnder
Günni
PCBs version 1.1 are on their way from China to Germany. Already ordered most parts for the Noiseunit, maybe i try the BC546/556 B Version instead of 2N3904/3906, with a Vceo of 65V. I have seen them often as differential input stage in poweramps. I will give them a try.
I have decided to go for 15V AC, since you pointed out that this design is not scalable to any AC voltage in the range of 12V to 18V AC. You also mentioned, that there are more effektive designs for the power class of 25W /8Ohms.
I think, that heatsinking at 15 V DC is still managable with your design. The availbale output power will be enough for my small office.
I found this site 404 Not Found for recalculation of the filter stages. So i can adopt the active filter to my needs.
Did you make any further changes in the design?
BR to DownUnder
Günni
Hi Günni,
The BC546/556 has a different pinout to the 2N3904/3906, so using them without PCB changes will be a real pain. I've gravitated towards the 2N5551/5401 for higher voltage parts, as they're interchangeable with the 2N3904 pinout wise but offer much higher voltage. If you really can't get the American transistors where you are then you could try MPSA06/56, which have the same pinout.
What impedance is your woofer? For the 4Ω woofer in noiseUnit I'm currently going with a 2x9V transformer - Farnell part number 9531769 (http://au.element14.com/multicomp/m...0va-2-x-9v/dp/9531769?st=toroidal transformer). That gives me supplies of +/-13.6V at ~2.8A (our mains supply here is about 250VAC, so is right on the high side of the standards), and drives a single 4" 4Ω driver hard into clipping with ease.
I'm also doing a new speaker for my room at work - tentatively called noiseUnit^2, which uses a pair of SB acoustics NRX12 4" 4Ω drivers in series, for a nominal impedance of 8Ω, and a single SB29RDNC tweeter. For that I'm going with a 2x12V transformer (Farnell 9531777), as the ~+/- 17VDC supply will better drive this speaker.
Also what are you using as a source? If you're plugging directly into a PC headphone socket or similar they'll be incredibly loud with the R1,2,10,11 resistor settings in the schematic. You can set whatever first stage gain you desire by selecting these four resistors.
You must post some pictures of your build!
The BC546/556 has a different pinout to the 2N3904/3906, so using them without PCB changes will be a real pain. I've gravitated towards the 2N5551/5401 for higher voltage parts, as they're interchangeable with the 2N3904 pinout wise but offer much higher voltage. If you really can't get the American transistors where you are then you could try MPSA06/56, which have the same pinout.
What impedance is your woofer? For the 4Ω woofer in noiseUnit I'm currently going with a 2x9V transformer - Farnell part number 9531769 (http://au.element14.com/multicomp/m...0va-2-x-9v/dp/9531769?st=toroidal transformer). That gives me supplies of +/-13.6V at ~2.8A (our mains supply here is about 250VAC, so is right on the high side of the standards), and drives a single 4" 4Ω driver hard into clipping with ease.
I'm also doing a new speaker for my room at work - tentatively called noiseUnit^2, which uses a pair of SB acoustics NRX12 4" 4Ω drivers in series, for a nominal impedance of 8Ω, and a single SB29RDNC tweeter. For that I'm going with a 2x12V transformer (Farnell 9531777), as the ~+/- 17VDC supply will better drive this speaker.
Also what are you using as a source? If you're plugging directly into a PC headphone socket or similar they'll be incredibly loud with the R1,2,10,11 resistor settings in the schematic. You can set whatever first stage gain you desire by selecting these four resistors.
You must post some pictures of your build!
Hello Suzy,
today arrived my PCBs for the noiseunit 1.1 from China.
Following questions:
- Do you solder the SMD parts manually, or do you have a reflow oven?
- which side has more Melf SMD parts (top or bottom); background is an article in a german computer magazine about soldering in an electrical pizza pan (SMD-Loten in der Pizzapfanne, Teil 2 |
Make) . Sorry it is in german language, but i think google transl. will help you
.
- my first attempt will be with NE5532, since i have about 20 pcs of them in my storage. Can i use OPA2134 for U3 since i have already here?
- for the first pcb i will use 2N3904 and 2N3906 (got it in Ammo Pack, so i need to bend back), because i have a pair of speakers in a Infinity Reference 1i case , with 5" Magnat woofer (4Ohm) and a Westra tweeter also 4 Ohm.
- the source will be a PGA2310/11 pre with input selector, often seen at ebay. So i can adopt the input level.
Still struggling to get all melf device, but only 4 values are missing, but i found two sources, one here in germnay (100pcs/ €1,65) and one in poland (50pcs/€1,20), 1% either from Beyschlag or Vishay.
No economical, to buy 100 pcs, when you need 4.
next step is fishing a pizza pan for 10€ in the bay.
I will report , when there is some progress
BR
Günni
today arrived my PCBs for the noiseunit 1.1 from China.
Following questions:
- Do you solder the SMD parts manually, or do you have a reflow oven?
- which side has more Melf SMD parts (top or bottom); background is an article in a german computer magazine about soldering in an electrical pizza pan (SMD-Loten in der Pizzapfanne, Teil 2 |
Make) . Sorry it is in german language, but i think google transl. will help you
.- my first attempt will be with NE5532, since i have about 20 pcs of them in my storage. Can i use OPA2134 for U3 since i have already here?
- for the first pcb i will use 2N3904 and 2N3906 (got it in Ammo Pack, so i need to bend back), because i have a pair of speakers in a Infinity Reference 1i case , with 5" Magnat woofer (4Ohm) and a Westra tweeter also 4 Ohm.
- the source will be a PGA2310/11 pre with input selector, often seen at ebay. So i can adopt the input level.
Still struggling to get all melf device, but only 4 values are missing, but i found two sources, one here in germnay (100pcs/ €1,65) and one in poland (50pcs/€1,20), 1% either from Beyschlag or Vishay.
No economical, to buy 100 pcs, when you need 4.
next step is fishing a pizza pan for 10€ in the bay.
I will report , when there is some progress
BR
Günni
Hi Günni,
I have access to a reflow oven (an Asscon Quicky 450) but find for simple boards like this it's easier to solder by hand.
There are roughly equal numbers of SMD parts top and bottom, so you'll probably do half by hand anyway if you're using a pan to reflow.
Almost all the resistors can be 1206 chip ones with no impact on performance. The only spots where you really have to be careful what resistors you use are where they're forming a voltage divider for negative feedback - I found with an earlier amp that thick film chip resistors introduce rather a lot of distortion when used for this job.
I used NE5532's originally with an OPA2132 for U3. This worked fine.
I have access to a reflow oven (an Asscon Quicky 450) but find for simple boards like this it's easier to solder by hand.
There are roughly equal numbers of SMD parts top and bottom, so you'll probably do half by hand anyway if you're using a pan to reflow.
Almost all the resistors can be 1206 chip ones with no impact on performance. The only spots where you really have to be careful what resistors you use are where they're forming a voltage divider for negative feedback - I found with an earlier amp that thick film chip resistors introduce rather a lot of distortion when used for this job.
I used NE5532's originally with an OPA2132 for U3. This worked fine.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.