T04 Gas Turbine - RESURRECTED!

[ashpowers]

In the pictures of the fuel ring the smaller injector "tips" are 1/8" o.d. and something like 0.012" wall. The reason they are so large is simply because I haven't installed the nozzles just yet. I was considering to take a different approach other than hypodermic needles. I have been thinking of making nozzles from some stainless steel that have a converging funnel on the inlet side and a diverging funnel on the outlet size, with a very shallow 0.027" bore between them - virtually just a knife-edge at the convergence of the inlet and outlet to this diameter of passage. This shape should produce a nice conical spray pattern to get the fuel onto the evap walls as best as possible.

I used 6 evaporators in this engine previously and it worked well. I think there were just issues with my combustor design that was causing the burn zone to be a little more centralized in the chamber rather than towards the primary zone. Unfortunately I cannot say with absolute certainty that this number of vaporizers is adequate or not though. I'll just have to build her and see how things work out.

[madpatty]

looking forward to see how you achieve that needle design....do posts some pictures if possible..

[ashpowers]

Been working on getting the new combustor put together over the past two days and it turned out an in acceptable format.

I'll be welding all the pieces together here in about an hour after I eat some lunch and then building a new fuel ring. The new ring design is slightly smaller in diameter due to the new vaporizer inlet location so gotta construct a new one. Probably have that done later this afternoon/evening and hoping to actually fire her off tonight for a few short test runs on propane.

[racket]

Hi Ash

Nice :-)

Cheers
John

[ashpowers]

Everything is welded together - took a good bit more time to get things right but I'm happy with the results. If this chamber works well I'll get my hands on some better metal that will last and make a new one - I already know a few things I would do a little differently in the fabrication that would make it turn out even better but for now this will work for experimentation.

Calling it a day but tomorrow I'll braze up the new fuel ring and should finally fire her up for the first time in about 6 years!

[rythmnbls]

Looking good Ash, that diffuser looks a lot like the kamps, lots of memories there.

Good luck on the first run.

Regards.

Steve.

[madpatty]

Hi Ash,
That thing is looking cool as it is being packed together...

Waiting for a video on her spoolup and a successful run...

Cheers,
Patty

[ashpowers]

So she's all assembled now and looking great. I built the new fuel ring and machined 3 small standoffs with small crossbores to fit a security wire through to wire tie the ring into plate.



I have held off on starting the engine as I wanted to address the electronics. I recently got my hands on a Samsung Galaxy Tab3 tablet (7") and really want to use this as my engine control unit rather than the Palm Tungsten PDA.


So I got to looking around for other DAQ solutions and came across this:
www.rocketmeter.com/index.html


This is BY FAR a better solution to engine control than what I have been using. This module has many IO ports, PWM output control, analog inputs, digital IO, and its completely wireless (802.11). It has reasonable sample rate, can control stepper motors, electric motors (we would need a MOSFET array for driving higher power motors, of course), and also has amperage/resistive IO ports.

This has everything we would need to construct a pretty robust control system. A few provisions would need to be made for signal conditioning, such as the tachometer and thermocouples but pressure sensors would tie directly in (with only a current limiting resistor for protection).

The unit also comes with the complete source code which will make it vastly more simple to modify it for our purposes. It was written in Basic4Android which I have and am familiar with. This is looking more and more like the direction I'm going to go. Best part is that one of these can be setup on multiple engines and the handheld device can be ported to any of them at any time. This will run on any android based handheld too - obviously the larger the screen the better the viewing, but there are loads of nice additional features that can be integrated into the program for our use.

Basic4Android has APIs for all of the integrated hardware onboard the handheld device. Such as the GPS (for speed and location), the accelerometers (which could be use to determine power based on a scalar of weight of the vehicle + pilot), and the camera - all of the embedded devices are available to use in the programming. Not to mention, all of the engine data can be recorded and reviewed at any time later as well as video, longitude/latitude recording to see your traveled path, etc etc - if you can imagine it, it's possible.

For the time being I am going to get the current control system all hooked up and packaged nicely so I can get to test running the engine. I want to apply a new tachometer arrangement that I commented on in another thread to the current system and work out any bugs in that while I am waiting to get my hands on the RocketMeter and source code. I'll have to build a small circuit that is all inclusive with the laser, phototransistor, schmitt trigger, and frequency to voltage converter so when the new hardware/software arrives I'll be ready to port that over.

All in all the RocketMeter costs $150 which is a really low price and I'll be making the turbine control program open source freeware for anyone who wants to get their hands on it and use it.

[ashpowers]

OK, so I've been playing with the laser tach circuit and have pretty much nailed it down.

This uses an LM741 opamp as a Schmitt trigger to convert the phototransistor signal to a digital signal with hysteresis to prevent false switching. Turn-on voltage is about 3.8V and turn-off voltage is about 2V. The output of the phototransistor is swinging from 4.4V all the way down into > 100mV when the beam is shining on it vs being blocked. Quite a significant swing which will be really good, especially on the high end as this gives the setup the best chance to work without sunlight affecting it. This will be fed into the lm2917N frequency to voltage converter circuit on the JECU (jet engine control unit). This voltage is linear to the rpm speed of the rotor group so a simple scalar value will be used to convert the voltage to an actual RPM value.

These two different images are showing the output state of the trigger based on if the beam is hitting the phototransistor or being blocked (which I did with a small screwdriver). This circuit should have good hysteresis to handle ambient lighting conditions and I will further test it once I have mounted everything to the turbine on the protoboard out in the sunlight. Once I've determined the best R1/R2/R3 resistor values that make this work without sunlight causing errors I'll build the circuitboard and put fixed value resistors. Then I'll condense this all down into a 1" X 0.5" circuit board to stuff under the inlet cowling on the engine.

[ashpowers]

Other good news!

I just heard back from the developer of the rocketmeter - I've got one of his units coming to me free of charge as of tomorrow! =) Freebies are always nice! He also said he is looking for some help, not sure exactly what that all entails but I see potential markets for the hardware he has, namely in my business with automotive technology - that piece of hardware can be used for so many applications and with all of the handheld devices out there today, tons of really nice features that can be built with them and this hardware. =)

So things are coming together for an all new FADEC system that should be pretty stout! I'll continue working on the current setup I have since it does work well and probably end up putting that onto my T25 turbine, which I also pulled out of storage this past weekend. =)

[ashpowers]

OK, so more development!

Schmitt trigger circuit. LM741: R1:10K, R2:6K, R3: 10K.
Trigger V: 3.64
Release V: 0.91

Phototransistor swing: .022V to 4.46V.
Will have to test this tomorrow in broad daylight to make sure the PT still drops below threshold but I think this setup will work just fine.



Laser, lasing.


So I got the idea how to test this configuration to make sure it will be able to properly read up to maximum RPM of the engine: 120,000RPM.

My dremel is rated up to 30,000RPM, not quite fast enough, however, with one of my diamond wheels having 8 holes just slightly larger than the laser beam, I'm in the money.

30,000RPM * 8 = 240,000. The nut on the compressor produces two pulses per revolution so 240,000 pulses/min. PERFECT. If this were setup as a reflective arrangement with only one pulse per rotation, this would support 240KRPM.

With the setup powered up, I put this wheel in the path of the beam starting at low speed and ramped it up to full speed - registering a nice smooth increase in pulse frequency up to 4.1KHz, or 123,000 RPM simulated, or 30,750RPM of the dremel motor. Even with small changes in the speed of the dremel running at full speed there was a distinct register of the pulse frequency variation so we are definitely good to go here.



I will be receiving a new USB cable for the Palm Tungsten PDA so I can make changes to the program to account for the changes to the tach circuit. From here it is looking really good to actually start her up tomorrow afternoon/evening! =)

Also, in my professional business many years ago I took the time to create a voltage to massflow conversion table for the Nissan 300ZX TwinTurbo mass airflow sensor. Once I get this engine up and running well I think I'll be taking it one step further and adding in this sensor to the setup to see what kind of massflow this engine is really moving. I think this info will be pretty useful to seeing if any further refinement should be made to the NGV and/or turbine to further optimize the performance of this engine.. That is, of course, if the rotorbearing group stays together long enough, LOL.

My biggest concern right now is with the rear bearings. They are deep-groove radial bearings, hybrids, phenolic retainer, ABEC#5 fit, 52100 chrome steel races. The races are limited in temp to ~220C. I purposely used angular contact bearings at the front end of the engine to carry the thrust loads since that area is cooler and just use a set of radial bearings in the rear that are press-fit to the shaft, back to back, and with a light preload spring. The interference fit is between 0.0002 and 00003mm. This fit definitely "sinched" up the slop in the bearings but without any significant drag. With them stacked back to back with this interference fit it is probably unnecessary to have a preload spring - the outer races do have a bit of drag between them - enough that only in the most brutal dump of lube into them would I expect the outer races to run at different speeds. But the preload spring is mostly in there to prevent the outer races from free-wheeling in the bearing tube - with a small side benefit to offsetting the axial loading put on the group by way of the pressure acting on the backside of the compressor wheel.

At the front end the two angular contact bearings are put back to back such that the front bearing butts up against the compressor diffuser and it carries the axial load. Its mate is oriented in opposing fashion and its outer race is preloaded by a spring washer to about 2lbs. This setup is mostly to put some rigidity into the front bearings to keep the shaft at center and isolating their preload from the rear bearings. The front bearing is preloaded primarily by the axial load by way of pressure behind the comp wheel, its mate is preloaded into it through the spring washer at the back face of its outer race. The front bearings constrain the axial position of the rotorgroup with only ~0.1mm of spring washer compression remaining in the rearward direction and clearances are in the parts to handle this without bind.

The front bearings are supplied with both cooling air and lube; a 1/4" diameter air "port" to feed into the bearing tunnel and a jetted feed of fuel/oil mix coming in from the fluid dampening sleeve area. The rear bearings also have a feed port to deliver a small spray of fuel/oil mix to them as well.

In the previous setup my cooling air flow area was considerably smaller than what I am using in this setup. The previous arrangement used 3 small ducts about 1mmX4mm between the bearing housing and the diffuser. Additionally I did not have a shaft seal behind the compressor wheel. My suspicion is that in large part, air coming in from behind the diffuser was actually passing through into the area behind the compressor wheel which is at lower pressure. The combined dilution of total pressure resulting from this probably reduced cooling air flow through the bearing tunnel back through the rear bearings and out behind the turbine wheel. This probably also added to less efficiency in the compressor section with the hot air that was recirculating from the combustor area into the area behind the compressor wheel and back out of the diffuser again. It only makes sense to have a seal behind the compressor to prevent this from happening as well as ensuring that the pressure at the front end of the bearing tunnel is maintained higher than the pressure behind the turbine wheel.

So even with all of this in play I'm not entirely certain that the rear bearings are going to hold up. From the reading I've done on this type of cage material I am looking at a peak operating temperature of around 120C (250F). I've come across documentation that puts this peak around 160C, still not much more though.

Its looking like I really should install a thermocouple to monitor this. There is room in the intermediary plate between the bearing housing and the NGV to place a sensor about 4-5mm behind the rear bearings. I have another 1/8" dia inconel sheathed K-type probe that is long enough to make a straight shot through the case, through the side of the NGV baseplate and on into this area just behind the rear bearings. I've also got another thermocouple signal conditioning circuit in the GT Control board so linking this in would be no problem.

After all the time spent putting this baby back together and all of the new upgrades she's had it would be a shame to see her come apart from an overheated/failed bearing. It is just really difficult to tell what kind of airflow is getting through the bearing tunnel. Fortunately in this setup, which I didn't previously mention, is the fact that the bearing lube is fed by an independent port at the front of the engine. I've ordered an inline ball-based flowmeter to put on this line so I can get an idea of what kind of fuel/oil feed is going into the bearing group. The line from the fuel pump is split so pressure to the lube system will follow fuel pressure. I am uncertain as to the sizing of the feed orifices in the bearing sleeve - right now they are pretty big, around 1mm in diameter. I know in the ballbearing turbochargers we use 0.035" restrictor jets for the oil feed but those systems operate at higher pressures - between 60-80psi. The fuel system in my setup, limited by the overpressure relief valve on the fuel pump, will bleed off at 80psi but there's also combustor pressure working against that, and I haven't seen any evidence that shows I was really anywhere near that kind of pressure.

I guess it is better to go more to the conservative side with a larger cooling air port and a little larger fuel/oil lube ports for the bearing group. However, too much lube to the bearings and you start having other problems too. Just playing around with a syringe to get an idea of fluid volume per time I speculate that the bearings would probably be happy with somewhere around 0.2 to 0.4ml/s so I'll try shooting for that to begin with and watching the bearing tube discharge temp.

I *may* get all of this together by tomorrow evening and have a chance to fire her off - just have to see how everything comes together through the day.

[racket]

Hi Ash

The air supply to the bearings can be a problem, when I first built the FM-1 engine I went for the RC micro turbine lube/cooling method with lube to the front bearing and air bleed from near the diffuser to shaft tunnel area ,( piston ring seal behind comp) but I found that the static pressure wasn't high enough to overcome the hot gas pressure at the other end of the shaft tunnel resulting in turb end bearing problems, I eventually fitted "snorkels" to take ram air from the diffuser outlets so that maximum static pressure was fed into the shaft tunnel.

Heres something that might throw some extra light on the subject , in recent correspondence with Smithy, who has an enormous amount of experience with RC micro engines, and who has recently experienced some regular turb end bearing life problems with a motor of his ,....... when the afterburner was ignited he felt there was a lack of air flow rearwards due to a "balancing" of air pressures , he's fitted force feeding lube supply to both bearings now and the problem has disappeared .

The cage material on your rear bearing might be a problem without a fair bit of air cooling , the bleed air will be pretty hot to start with , possibly near 100 deg C at full power, it doesn't leave much of a buffer for the temperature rise generated by simply spinning around at extreme rpms.

Maybe also check that your 1/4" air port isn't being subjected to high speed air flowing across it and creating a "suction" , not the right way to describe it , but think of fuel exiting into a carburetor throat where the high speed air creates the suction.

Looking forward to seeing the TO4 burning kero :-)

Cheers
John

[ashpowers]

Well, I successfully installed a K-type thermocouple probe, 1/8" diameter, inconel sheath, straight through the casing, through the NGV base and right into the space just behind the rear bearings. The tip is not contacting any part of the engine and comes in just a few mm radially from the turbine shaft. Unfortunately it really isn't easily possible to have this sensor contacting any part of the bearing - the inner race is obviously spinning and the outer race is contained within the bearing sleeve that has oring seals at each end. AT least I will be able to get a measure on the temperature of the air/fuel/oil mix that is coming out of the bearing sleeve.

I have completed the modifications to the programming as well in the Palm PDA - tested and calibrated the tachometer and setup the additional thermocouple circut input for the bearing group with temp display on the screen. I'll post pics later today once I have everything fully setup and ready for bench testing.

I have my fingers crossed that the combination of the good quantity of cooling air combined with the fact that the fuel/oil mix is completely enveloping the bearing sleeve to produce the fluid damping arrangement AND both front and rear bearing sets have their own fuel/oil feed that this baby will survive the test of time. In the previous arrangement the balance was horrible, single bearings at each end, cheap bearings, no fluid dampening, limited cooling air AND fuel/oil lube feed, and no compressor shaft seal. Also no means to monitor bearing tunnel temps. Hoping that this "shotgun" approach is a success today but if bearing tunnel temps get into the 150C's or so I will shut her down and seek out some M50 bearings or possibly full ceramic for the rear end and proceed from there.

As for the cooling air flow - the entry port was milled with the center of the bore right at the interface plane between the compressor diffuser and the bearing housing. Once it was punched through, I went through a series of larger drill bits to make a bit of a velocity stack shape leading into this port. Just inside the port I used a ball shaped carbide to radius the port to get the airflow moving down the tunnel. It all came together pretty well and I'm happy with it so we will just have to see. I do have a couple of other ideas on how to mitigate the temps in the bearing system if this doesn't pan out though.

[ashpowers]

So I tried starting the old girl up this afternoon. Too many problems trying to feed both propane and jet fuel into the same orifice. More tubing will be here tomorrow so I can install a proper propane feed line to one of the vaporizers. The bearing tunnel sensor at one point did peak up to around 200C but I could never get the engine running to make sufficient P2 pressure. Ended up pumping some fuel/oil into the bearing tunnel feed line followed by some good long air blasts into the feed line to cool things down. The rotating group spins very freely with no noises - just that smooth sound they are supposed to make. I pulled the motor apart to take a look at things and everything looks great other than a hotspot on the outer liner of the combustor in about the 3-o'clock position. I'm thinking that I never got the engine up to self sustain speeds and less than 1psi of p2 pressure and with several failed start attempts where it dumped a chunk of fuel into the chamber and big flames coming out the back while I quicly covered the inlet and blasted air into the jetpipe to snuff out the fire before it did any damage. Re-tries after that probably had small pools of fuel in the engine that were burning off and licking flames onto the combustor liner and left the hotspot I see. I'll have to reinspect her once I have a chance to run her up to full speed. No successful start but the solution to this is an easy one.

Will try again tomorrow.

(Its been a while since I have had this smell in my garage. Puts a smile on my face!)

[madpatty]

Hi Ash,
Maybe you have mentioned somewhere in the thread...

What are you using to spool up the engine to self substaining rpm's?

Right now i am in the quest to build an electric starter for my engine because the blower i used to spool my earlier engine is not able to spool this one....

Cheers,
Patty